Author + information
- Received January 7, 2016
- Revision received May 9, 2016
- Accepted May 15, 2016
- Published online August 8, 2016.
- Carlos Cortés, MDa,
- Ignacio J. Amat-Santos, MD, PhDa,∗ (, )
- Luis Nombela-Franco, MDb,
- Antonio J. Muñoz-Garcia, MDc,
- Enrique Gutiérrez-Ibanes, MDd,
- José M. De La Torre Hernandez, MD, PhDe,
- Juan G. Córdoba-Soriano, MDf,
- Pilar Jimenez-Quevedo, MDb,
- José M. Hernández-García, MD, PhDc,
- Ana Gonzalez-Mansilla, MDd,
- Javier Ruano, MD, PhDe,
- Jesús Jimenez-Mazuecos, MD, PhDf,
- Javier Castrodeza, MDa,
- Javier Tobar, MDa,
- Fabian Islas, MDb,
- Ana Revilla, MDa,
- Rishi Puri, MBBS, PhDg,
- Ana Puerto, MSca,
- Itziar Gómez, MSca,
- Josep Rodés-Cabau, MD, PhDg and
- José A. San Román, MD, PhDa
- aInstitute of Heart Sciences, Hospital Clínico Universitario, Valladolid, Spain
- bHospital Clínico Universitario San Carlos, Madrid, Spain
- cHospital Clínico Virgen de la Victoria, Málaga, Spain
- dHospital General Universitario Gregorio Marañón, Madrid, Spain
- eHospital Universitario Marques de Valdecilla, Santander, Spain
- fHospital General Universitario de Albacete, Albacete, Spain
- gQuebec Heart and Lung Institute, Quebec, Canada
- ↵∗Reprint requests and correspondence:
Dr. Ignacio J. Amat-Santos, Institute of Heart Sciences, Hospital Clínico Universitario, Ramón y Cajal 3, 47005 Valladolid, Spain.
Objectives This study sought to analyze the clinical impact of the degree and improvement of mitral regurgitation in TAVR recipients, validate the main imaging determinants of this improvement, and assess the potential candidates for double valve repair with percutaneous techniques.
Background Many patients with severe aortic stenosis present with concomitant mitral regurgitation (MR). Cardiac imaging plays a key role in identifying prognostic factors of MR persistence after transcatheter aortic valve replacement (TAVR) and for planning its treatment.
Methods A total of 1,110 patients with severe aortic stenosis from 6 centers who underwent TAVR were included. In-hospital to 6-month follow-up clinical outcomes according to the degree of baseline MR were evaluated. Off-line analysis of echocardiographic and multidetector computed tomography images was performed to determine predictors of improvement, clinical outcomes, and potential percutaneous alternatives to treat persistent MR.
Results Compared with patients without significant pre-TAVR MR, 177 patients (16%) presented with significant pre-TAVR MR, experiencing a 3-fold increase in 6-month mortality (35.0% vs. 10.2%; p < 0.001). After TAVR, the degree of MR improved in 60% of them. A mitral annular diameter of >35.5 mm (odds ratio: 9.0; 95% confidence interval: 3.2 to 25.3; p < 0.001) and calcification of the mitral apparatus by multidetector computed tomography (odds ratio: 11.2; 95% confidence interval: 4.03 to 31.3; p < 0.001) were independent predictors of persistent MR. At least 14 patients (1.3% of the entire cohort, 13.1% of patients with persistent MR) met criteria for percutaneous mitral repair with either MitraClip (9.3%) or a balloon-expandable valve (3.8%).
Conclusions Significant MR is not uncommon in TAVR recipients and associates with greater mortality. In more than one-half of patients, the degree of MR improves after TAVR, which can be predicted by characterizing the mitral apparatus with multidetector computed tomography. According to standardized imaging criteria, at least 1 in 10 patients whose MR persists after TAVR could benefit from percutaneous mitral procedures, and even more could be treated with MitraClip after dedicated pre-imaging evaluation.
Aortic stenosis is the most frequently treated valvular disease in Western countries, followed by mitral regurgitation (MR) (1). Degeneration is considered the main etiology for dual valve dysfunction, with the prevalence of aortic stenosis increasing by approximately 2% per year beyond 75 years of age (2). Age and associated comorbidities frequently determine the mode of valve treatment, with up to two-thirds of AS patients over the age of 75 years now undergoing transcatheter aortic valve replacement (TAVR) (3–6). Clinical guidelines recommend double valve intervention if both valves are severely diseased; nevertheless, there is currently a dearth of evidence supporting this strategy (7,8). Although the Euro Heart Survey reported a 2-fold increase in mortality after double valve surgery (1), recent investigations suggest a benefit derived from treating both valves (9).
In the TAVR arena, the impact of concomitant MR on clinical outcomes has not been evaluated formally across the main TAVR trials given that severe MR has been frequently considered an exclusion criteria (4–6). Whereas the presence of significant MR at baseline has been linked previously with greater mortality, it is not clear whether its improvement implies a reduction in mortality (10,11). Transthoracic echocardiography (TTE) and transesophageal echocardiography (TEE) and, increasingly, multidetector computed tomography (MDCT) are used to characterize the mitral apparatus (10); however, the severity and accurate etiological assessment of MR is often challenging, with a lack of a formal validation of the imaging predictors of improvement (10–12). Careful imaging characterization of the mitral valve is crucial for determining feasibility of potential percutaneous techniques for mitral valve disease. We analyzed the clinical impact (13) of the degree and improvement of MR in TAVR recipients, and validated the main TTE/TEE and MDCT determinants of this improvement using central off-line imaging analysis (12). We further assessed the proportion of potential candidates for double valve repair with currently available percutaneous techniques.
Between August 2007 and January 2015, 1,110 consecutive patients without previous mitral prostheses and with a concomitant diagnosis of severe aortic stenosis underwent TAVR across 6 centers. Patients were classified into 2 groups according to the degree of baseline MR following the European Guidelines of Echocardiography (7,8): 1) low-degree MR for those with grades 0 (none), 1 (mild), and 2 (nonsignificant moderate MR); and 2) high-degree MR for grades 3 (significant moderate MR) and 4 (severe MR). For statistical purposes, any decrease of 1 or more degrees was considered an improvement of the MR severity.
All patients underwent complete 2-dimensional and color Doppler echocardiography at baseline, before discharge, and at 1 and 6 months after TAVR. Central off-line analysis of all images was performed by an independent core lab (ICICORELAB, Valladolid, Spain), which collected and interpreted all imaging data blinded to the clinical data and outcome. Baseline measurements were graded following the recommendations of the European Association of Echocardiography (14). Analysis of the degree and etiology of MR was performed by an expert in echocardiography blinded to further data of the post-TAVR outcomes as well as to MDCT measurements. Other parameters included end-diastolic and end-systolic left ventricular diameters in the long axis parasternal view, left ventricular ejection fraction (Simpson's method), determination of the organic or functional/ischemic origin of MR, mitral and aortic annular diameters, and tenting height and area when indicated. The mitral valve annulus was measured using the mid-diastolic 4-chamber and 2-chamber views, and tenting distance was measured as the distance between coaptation point of mitral leaflets and annular plane of mitral valve using the mid-systolic parasternal long axis view. Tenting area was measured as an area enclosed between leaflets and annular plane.
MDCT analysis was performed in most patients as part of the standard protocol to evaluate cardiac structures and the aorto–iliac–femoral axis before transcatheter aortic valve implantation. Off-line analysis was carried out only for those patients with significant MR (at any time point) with available MDCT images (87%). Across all centers, MDCT exams were performed with a 64-row scanner during inspiratory breath-hold using iodine contrast (Omnipaque 350 mg 1/ml; volume adjusted depending on weight and renal function). Scans were reconstructed with a section thickness of <1 mm. Apart from standard measures (i.e., aortic annulus dimensions, coronary ostia height), MDCT allowed the characterization of the mitral apparatus in patients with significant MR, including mitral annular diameters in 4- and 2-chamber-views, and location/degree of calcification affecting the mitral annulus and leaflets according to a semiquantitative score previously described (12) (Online Table 1). Also, the presence or absence of calcification at the mitroaortic junction was assessed.
Finally, we assessed the proportion of potential candidates for currently available percutaneous mitral therapies which included edge-to-edge repair with the MitraClip device (Abbott Vascular, Santa Clara, California) (15), and balloon-expandable valve implantation within a severely calcified mitral annulus for patients with organic MR (16), according to previously reported criteria that have been summarized in Online Table 2.
Data are expressed as absolute frequency and percentages for qualitative variables. Quantitative variables are described as mean ± SD or median [25th to 75th interquartile range], depending on variable distribution. Group comparisons according to the degree and improvement of MR were analyzed using Student t test or its nonparametric equivalent, Mann-Whitney U test for continuous variables, and chi-square test or Fisher exact test for categorical variables. Those variables exhibiting a p value of <0.10 in the univariate analysis were included in a multivariate analysis to determine independent predictors of 6-month global mortality and MR improvement was performed. Survival curves for 6-month global and cardiac mortality were compared with a log-rank test according to MR severity. Receiver operating characteristic curve analysis was performed to discriminate the utility of the mitral annulus diameter (as measured in 4-chamber view) for predicting MR improvement. Finally, the intraobserver variability was determined for mitral annulus diameter (as measured by echocardiography and MDCT; repeated measures 1 month apart for 20% of the patients) through intraclass correlation coefficient and for degree of calcification of mitral apparatus through class ponder Kappa for both, MDCT and echocardiography; the strength of agreement was described according to the scale by Landis and Koch (17). Because many comparisons were performed, p < 0.01 was used as cutoff for statistical significance. All analyses were conducted using the statistical package SPSS, version 20.0 (SPSS, Inc., Chicago, Illinois).
Clinical and imaging baseline characteristics of the study population
The mean age of the global TAVR population (n = 1,110) was 80.5 ± 6.9 years and 465 (41.9%) were men. Main clinical and imaging characteristics of the global population according to the baseline degree of MR are summarized in Table 1. A total of 177 patients (15.9%) presented with significant MR (≥3) at baseline evaluation by TTE and up to 43.3% (n = 480) presented with at least moderate MR (≥2).
As shown in Table 1, despite a comparable left ventricular ejection fraction, mean transaortic gradients at rest were higher and indexed aortic valve area lower in patients with significant MR. However, pulmonary artery systolic pressure was comparable between both groups. Patients with significant MR most likely had a functional etiology than those with MR of low-degree (37.3% vs. 27.5%; p = 0.009), and also larger mitral ring (34.5 ± 4.0 mm vs. 32.7 ± 4.8 mm of diameter; p = 0.001) and tenting parameters. Thirty-seven percent of the patients with significant MR underwent TEE. Of those initially considered as harboring functional MR, 72.4% were reclassified to the organic MR category, whereas 13.5% of those initially thought to harbor organic MR were reclassified to the functional category.
After evaluation of mitral apparatus calcification by both TTE and MDCT, intraobserver variability values for ponder Kappa were 0.980 (95% confidence interval [CI]: 0.965 to 0.996) and 0.960 (95% CI: 0.935 to 0.985), respectively. Intraclass correlation coefficients for the measures of mitral annular diameters (4-chamber view) by both techniques were 0.993 (95% CI: 0.992 to 0.995) for TTE and 0.974 (95% CI: 0.966 to 0.981) for MDCT.
Clinical and imaging predictors of MR improvement in patients with significant MR at baseline
A total of 103 patients (58.2%) of 177 with significant baseline MR demonstrated an improvement of at least 1° after TAVR (Figure 1). The main clinical, procedural, and imaging predictors of this improvement are summarized in Online Table 3. The use of different types of valves impacted mortality, but not the evolution of MR severity. Rhythm (new-onset atrial fibrillation [NOAF]) and conduction disturbances (new left bundle branch block or the need for permanent pacemaker) were related to persistence of MR degree.
Also, several imaging characteristics were related to the evolution of MR (Online Table 3). In particular, diagnosis of organic MR by TTE was associated with persistence of MR; however, in multivariate analysis, the subjective evaluation of MR as being organic or functional by TTE failed to predict whether MR changed. However, calcification of the mitral apparatus by MDCT was found to be the optimal parameter for predicting MR improvement post-TAVR, together with mitral annular size (with a cutoff value of 35.5 mm as the diameter that best predicted MR evolution) (Figure 2) and NOAF (Table 2).
Potential candidates for percutaneous mitral therapies among TAVR recipients
A flow chart with the selection of the patients who could benefit from percutaneous mitral valve repair according to the criteria described in the Methods (Online Table 2) is shown in Figure 3. Up to 14 patients (1.3% of all transcatheter aortic valve implantation patients) were deemed to be potentially suitable for some of the percutaneous mitral therapies currently available, representing 13.1% of all those patients with persistent MR ≥3 at 1 month after TAVR (n = 107). Specifically, 4 patients (3.8%) with organic MR were deemed to benefit potentially from balloon-expandable valve implantation within a mitral annulus with circumferential calcification and area as measured by MDCT below 720 mm2 (Figures 4A and 4B), whereas 10 patients (9.3%) with significant MR presented with criteria deemed appropriate for MitraClip implantation (Figures 4C to 4E). Within this last group, 8 patients presented with functional MR (80%) and 2 with isolated prolapsed leaflet segment (20% of organic MR patients). None of these patients were in New York heart Association functional class I at the 1-month follow-up, and 12 of them (85.7%) had died at 6 months after TAVR.
Procedural, in-hospital, and follow-up outcomes of the study population
Main procedural and follow-up outcomes are reported in Online Table 4. Compared with those without significant MR, those with significant MR were more commonly treated via the transfemoral approach (88.7% vs. 82.5%; p = 0.046), underwent balloon pre-dilation (96.2% vs. 78%; p < 0.001), and received a self-expandable valve (87% vs. 78.2%; p = 0.008). The procedure was considered successful in 95% of the patients without differences according to the baseline degree of MR. However, the need for hemodynamic support was more frequent when significant MR was present than in patients with lower degrees of MR (9.5% vs. 2.9%; p < 0.001).
Median hospital stay was 6 (4 to 9) days with longer duration among patients with concomitant MR of ≥3 compared with those without significant MR (9 [7 to 14] days vs. 5 [4 to 7]; p < 0.001). Residual aortic regurgitation was more common in those with significant MR (10.5% vs. 1.8%; p < 0.001), with comparable rates of post-dilation but a higher rate of conduction disturbances, leading to the need for permanent pacemaker implantation in 31.4% of the patients with significant MR as opposed to 23.2% in the group with an MR of <3 (p = 0.022). The presence of NOAF during the hospitalization was found in 17.9% of the patients with double valve disease, in contrast with 9.8% in the group with isolated aortic valve disease (p = 0.003).
Finally, in-hospital and 6-month mortality rates were 5.3% and 14.1%, respectively, with differences noted at both time points according to the degree of MR as depicted in Figure 5A.
Predictors of follow-up mortality
The main predictors of 6-month mortality post-TAVR are summarized in Table 3. Of note, 60% of the 6-month mortality was considered cardiac in nature according to Valve Academic Research Consortium (VARC-2) criteria (13). Patients with significant MR at baseline had a 3-fold increase in all-cause mortality compared with those without significant baseline MR (35.0% vs. 10.2%; p < 0.001), which was also higher if only cardiac mortality was considered (20.9% vs. 6.0%; p < 0.001). No difference in all-cause mortality during follow-up was found in the subgroup of patients with baseline significant MR.
This multicenter study involving a centralized core laboratory analysis of TTE/TEE and MDCT imaging demonstrated the presence of concomitant significant baseline MR in 16% of TAVR recipients. The presence of double valve (aortic and mitral) disease associated with greater mortality, yet improvement in MR (demonstrated in approximately 60% of individuals) after TAVR did not translate into improvements in short- and mid-term all-cause mortality despite a reduction in cardiac deaths. Only 1.3% of the global TAVR population were deemed potential candidates for currently available percutaneous mitral valve therapies, including MitraClip and balloon-expandable prostheses, for those with severely calcified mitral apparatus. This analysis thus comprises the first systematic description of MDCT parameters that could aid in the prediction of MR persistence after TAVR, further assisting the ability to characterize the optimal therapeutic approach (percutaneous or surgical) for treating concomitant residual significant mitral valve disease in these TAVR recipients.
Prognostic impact of MR
The presence of significant MR at baseline associated with a >3-fold increase in mortality, with a 35% incidence at the 6-month post-TAVR follow-up; 60% of deaths were cardiac related. This was surprising given that, in our population, the baseline risk estimated by the Logistic EuroSCORE was lower in patients with a baseline MR of ≥3, suggesting that the impact of multivalvular disease may be greater than expected. Several factors may have influenced these poor outcomes; however, it seems that more advanced fibrous skeleton degeneration and greater calcification as determined by MDCT, coupled with larger dimensions of the mitral apparatus (Table 1), likely contributed to more frequent conduction system disturbances in patients with an MR of ≥3. Calcification of cardiac structures is well-recognized as a predictor of earlier mortality (18,19). Moreover, greater degrees of calcification of cardiac structures may be the substrate for the higher rate of residual aortic regurgitation, which is known to associate with greater mortality (20). The same complications (conduction disturbances and aortic regurgitation) occur more often after self-expandable valves (10), which could explain the higher mortality of these patients when compared with those treated with balloon-expandable devices.
Mixed association between MR and post-TAVR outcomes had been reported in previous series (21). Noncardiac deaths are probably related to the baseline burden of medical comorbidities and are unlikely to be ameliorated despite improving the grade of MR. Hence, differences in the proportion of noncardiac mortality rates across different series (4–6,22–24) may confound the true benefit of improving MR in certain TAVR recipients. In our series, the improvement of baseline MR failed to improve the rate of cardiac mortality; nevertheless, there was a tendency to higher cardiac mortality in patients with a persistent MR of ≥3 (24.4%) than in those who had a decrease in the MR degree to <3 (15.7%; p = 0.151). Moreover, cardiac mortality rate was 3 times higher in patients with an MR of 4 (45.2%) than in those with an MR of 3 (15.7%; p < 0.001). According to these results, there is a potential for improvement in outcomes (through reduction in cardiac mortality) if persistent MR is treated, especially in those with higher degrees of residual MR.
Predictors of MR improvement
The impact of NOAF on MR persistence after TAVR has been described previously (21,25), representing perhaps a marker of more advanced cardiac disease (26). The impact of various technical aspects as aortic valve pre-dilation and the type of valve implanted remains controversial (21,25). What seems crucial is a detailed description of the cardiac anatomy, calcific burden, and its distribution on MDCT. Its presence in moderate to severe degrees within the mitral annulus and leaflets associates with greater stiffness of mitral structures, suggesting an organic nature of mitral valve disease, thus hindering MR improvement after TAVR. If this high degree of calcification appears together with a larger annulus diameter (which has been suggested to occur often in primary MR) (27), it may preclude from regression of the dilation despite correction of the ventricular pressure overload. On the contrary, greater calcification of the aortic valve reflects a more severe aortic stenosis whose correction may diminish more left ventricular pressure overload than in patients with less calcified aortic valves. This may contribute subsequently to lowering the MR degree in patients with more calcified aortic valves. This apparent paradox suggests that precise location of cardiac calcium could be more important than the mere quantification of noncoronary cardiac calcification. Therefore, a thorough evaluation of these imaging parameters could have important clinical relevance when planning a therapeutic strategy for high surgical risk patients with multivalvular disease.
Double valve surgery versus double percutaneous valve therapies
Answering the question regarding which high surgical risk patients should undergo double valve surgical repair or replacement instead of TAVR remains challenging and fraught with caveats. The worldwide enthusiastic embrace of TAVR, leading to its use in lower surgical risk patients, perhaps makes this question even more relevant. Moreover, the addition of new percutaneous alternatives for treating mitral valve disease increases the complexity of the decision tree. The probability of MR persistence after TAVR in patients without atrial fibrillation, pulmonary hypertension, severe mitral apparatus calcification, and mitral annulus diameter of <35.5 mm was rather low (<10%). However, there is a wide range of combinations of these factors. Only 14% of patients with minimal or absent mitral annular calcification on MDCT demonstrated persistently greater degrees of MR, suggesting this to be a dominant factor in the search for a marker of MR evolution after TAVR.
Recently, some investigators have suggested that double valve surgery may be related with better clinical outcomes (10), contradicting previously suggested treatment paradigms (28). Indeed, in the substudy from PARTNER (Placement of Aortic Transcatheter Valve) trial by Barbanti et al. (29), patients who were operated and presented concomitant moderate to severe MR had a 2-fold increase in mortality. On the contrary, mortality remained similar after TAVR irrespective of the MR degree, although severe MR was scarce given that it was an exclusion criterion. Some other TAVR studies have suggested isolated aortic valve treatment as the optimal approach (25), even though, in these studies, between 15% and 60% of patients demonstrated residual significant MR (10). Only 1 series has reported systematic percutaneous treatment of the mitral valve after TAVR (Online Ref. 1), demonstrating adequate clinical outcomes at the 1-year follow-up. Some case reports (Online Refs. 2–7) have presented successful balloon-expandable prosthesis implantation in mitral position after or together with TAVR. However, none of these have reported the potential number of patients who may benefit from these innovative strategies, and in which cases would combined percutaneous and surgical techniques be required, carrying important clinical and cost-effectiveness implications. The low rate of patients who may benefit, however, from double valve percutaneous therapies according to our results (<1.5%), is in contrast with the approximately 10% of patients with aortic stenosis who undergo concomitant surgical aortic and mitral valve replacement or repair (Online Ref. 8), suggesting the need for a more accurate patient-level therapeutic decision making process. Although the estimation of candidates to balloon-expandable valve implantation in mitral position was probably accurate, the number of potential candidates to MitraClip may probably be higher if dedicated evaluation were prospectively performed. Indeed, the proportion of patients that were excluded for both TAVR and surgery due to concomitant MR remains undescribed.
The main limitation of the present analysis pertains to its retrospective nature, whereby case collection and exclusion of some patients deemed surgical or inoperable could have occurred. Future studies in larger populations with longer and prospective follow-up are needed to clarify the actual impact on MR reduction and mortality for TAVR recipients with concomitant MR.
MR is not uncommon in TAVR recipients, driving subsequent clinical outcomes. More than one-half of these patients will improve after TAVR, with an associated reduction in the proportion of cardiac mortality. Several imaging parameters obtainable from echocardiography and MDCT may help to identify those patients whose MR is unlikely to improve after TAVR. In addition, both techniques can help to identify the relatively small proportion of patients who may benefit from concomitant or staged aortic and mitral percutaneous procedures.
WHAT IS KNOWN? Nearly one-half of the patients who present mitral regurgitation had an improvement of its degree after transcatheter aortic valve replacement.
WHAT IS NEW? Computed tomography can be useful to identify who will improve and to determine which patients treated with transcatheter aortic valve replacement can benefit from percutaneous mitral valve therapies.
WHAT IS NEXT? The high rate of multivalvular disease requires comprehensive innovative solutions that are warranted in the near future.
For an expanded reference section and supplemental tables, please see the online version of this article.
Project granted by the Instituto de Salud Carlos III (PI14/00022). Dr. Rodés-Cabau has received research grants from Edwards Lifesciences, St. Jude Medical, and Medtronic. Dr. Hernández-García has received research grants from Medtronic.
All other authors have reported that they have no relationships relevant to the contents of this paper to disclose. Drs. Cortés and Amat-Santos contributed equally to this work.
- Abbreviations and Acronyms
- multidetector computed tomography
- mitral regurgitation
- new-onset atrial fibrillation
- transcatheter aortic valve replacement
- transesophageal echocardiography
- transthoracic echocardiography
- Received January 7, 2016.
- Revision received May 9, 2016.
- Accepted May 15, 2016.
- American College of Cardiology Foundation
- Iung B.,
- Baron G.,
- Butchart E.G.,
- et al.
- Lindroos M.,
- Kupari M.,
- Heikkilä J.,
- Tilvis R.
- Iung B.,
- Cachier A.,
- Baron G.,
- et al.
- Nishimura R.,
- Otto C.M.,
- Bonow R.O.,
- et al.
- Vahanian A.,
- Alfieri O.,
- Andreotti F.,
- et al.
- Harling L.,
- Saso S.,
- Jarral O.,
- et al.
- Nombela-Franco L.,
- Ribeiro H.B.,
- Urena M.,
- et al.
- Amat-Santos I.J.,
- Revilla A.,
- López J.,
- et al.
- Lancellotti P.,
- Tribouilloy C.,
- Hagendorff A.,
- et al.
- Clavel M.A.,
- Pibarot P.,
- Messika-Zeitoun D.,
- et al.
- Nombela-Franco L.,
- Eltchaninoff H.,
- Zahn R.,
- et al.
- Rodés-Cabau J.,
- Webb J.G.,
- Cheung A.,
- et al.
- Thomas M.,
- Schymik G.,
- Walther T.,
- et al.
- Naoum C.,
- Leipsic J.,
- Cheung A.,
- et al.
- Barbanti M.,
- Webb J.G.,
- Hahn R.T.,
- et al.