Background and aim of the study: Several in vitro studies have reported on cavitation bubble visualization with mechanical heart valves and the cavitation intensity has been correlated with the design of the valve, the load on the valve occluder, the velocity of the occluder tip, and the negative pressure transients in the vicinity of the occluder. These studies demonstrated the presence of cavitation for certain types of valves under simulated normal physiological loading conditions. However, extrapolation of these results to cavitation initiation in vivo has been questioned due to limitations of the in vitro studies in simulating the in vivo tissue compliance. The present study was intended to analyze the effect of valve holder flexibility (simulating compliance of the suture ring and the surrounding tissue in vivo) on cavitation dynamics.

Methods: Cavitation bubbles were visualized on three types of mechanical heart valves (Medtronic Hall, Edwards-Duromedics, and CarboMedics) in our in vitro set up, and pressure transients were measured close to the occluder at valve closure. Two different flexible valve holders made of Teflon (elastic modulus, E = 400 MPa) and low density polyethylene (E = 180 MPa) were employed and the results were compared with those with a rigid Plexiglas holder (E = 2930 MPa) of the same geometry (3'' by 3'' wide and 1/8'' thick).

Results: Significant reductions were noted in the intensity of cavitation bubbles appearing along the clearance region of the Medtronic Hall and CarboMedics valves, with increasing valve holder flexibility. However, no attenuation was observed for the bubbles appearing around the seating lip or stop of the Edwards-Duromedics and Medtronic Hall valves that are believed to be caused by the fluid squeezing effect.

Conclusions: The results of the study suggest that timing of the mechanism to initiate cavitation is a critical factor in cavitation attenuation with flexible valve holders. If cavitation is initiated before the flexible valve holder responds to the impact at valve closure (such as due to squeeze film effect), cavitation intensity remains unchanged. Based on the results of the study, we propose that tissue compliance in vivo may not attenuate cavitation initiation for certain types of mechanical heart valves depending on the cavitation initiation mechanism.

How to cite: Lee, C. S., Aluri, S., & Chandran, K. B. (1996). Effect of valve holder flexibility on cavitation initiation with mechanical heart valve prostheses: an in vitro study. The Journal of heart valve disease, 5(1), 104–113.

Background and aim of the study: The squeeze-flow that develops during valve closure is believed to cause cavitation in mitral mechanical heart valves (MHVs).

Methods: Squeeze-flow was studied in tilting disc MHVs using two different numerical approaches. In the first decoupled analysis, experimental measurements of valve closing velocities were input into a computer model which simulated the resulting flow field. The second coupled approach involved simulation of the occluder motion and housing deformation in response to the surrounding flow.

Results: Both models predicted the likelihood of cavitation during the squeeze-flow phase of valve closure. They also indicated that valve mounting compliance influences the squeeze-flow field. The coupled analysis also showed that squeeze-flow is influenced by fluid viscosity and the geometry of the contact region.

Conclusions: These parameters could therefore influence the inception of cavitation in tilting disc mitral MHVs.

How to cite: Makhijani, V. B., Siegel, J. M., Jr, & Hwang, N. H. (1996). Numerical study of squeeze-flow in tilting disc mechanical heart valves. The Journal of heart valve disease, 5(1), 97–103.

Background and aim of the study: Transient cavitation has been directly observed near operating mechanical heart valves in vitro and inferred in vivo via the observation of pitting on explanted clinically used valves. Visual detection of cavitation bubbles, however, cannot be accomplished in vivo or when any opaque fluid, e.g. blood, is used.

Methods: This study examines a passive acoustic technique for detecting cavitation caused by a 27 mm tilting disc valve. We captured, valve closing sounds in vitro and attempted to detect a shift of energy into higher frequencies due to emission of broad-band noise caused by collapsing bubbles. The valve tester consists of a piston pump which directly drives the valve, a 16 cm diameter cylindrical lucite atrium and an air chamber in parallel to provide compliance. Water was used as a blood analog fluid. The cycle rate was altered to vary valve loading and produce cavitation. Acoustic signals were detected by a miniature hydrophone and a large area transducer and the waveforms were spectrum analyzed to 200 kHz and 500 kHz respectively. Cavitation onset was determined rising a high speed video camera.

Results: It was found that even under non-cavitating conditions, significant energy was produced at frequencies greater than 100 kHz, and this energy increased with increased load. The proportion of energy in high frequency bands, however, remained fairly constant when cavitation was not present and began to rise only after the cavitation threshold was reached. To isolate cavitation as an independent variable, data were taken with all system parameters constant, but using water under two different conditions. Degassed 17 degrees C water produced no visualizable bubbles, while aerated tap water at 43 degrees C showed a high degree of cavitation.

Conclusions: The results indicate that cavitation, while causing a shift of energy to higher frequencies, is not the only mechanism responsible for the shift of energy into higher frequencies.

How to cite: Herman, B. A., Porter, J. M., & Carey, R. F. (1996). Study of an acoustic technique to detect cavitation produced by a tilting disc valve. The Journal of heart valve disease, 5(1), 90–96.

Background and aim of the study: In clinical research, survival, reliability and failure analyses, the use of censored lifetime data often becomes a necessity. In this paper we present a novel methodology developed to allow for the use of censored data to train neural networks to predict the time of specific adverse events.

Methods and results: Specifically, for patients with implanted bioprostheses, we were able to design and train a neural system to successfully predict the time from valve implant to valve dysfunction. Further, we were able to demonstrate the clear improvement in performance and predictive accuracy of the system when trained using this method. The assertion that censored data carry additional and extremely valuable information, especially in cases of rare events, is substantiated by this correlation analysis.

Conclusions: This new methodology, in combination with results obtained from previous models which were able to identify the patients most likely to experience such events, now completes the picture by pin-pointing the 'who', as well as the 'when'.

How to cite: Katz, A. S., Katz, S., & van der Wal, E. A. (1996). Censored lifetime data in adaptive neural networks. The Journal of heart valve disease, 5(1), 84–89.

Background and aim of the study: Thrombosis remains a serious risk for patients with artificial heart valves and may be attributed in part to adverse blood flow patterns. Although the final assessment of a valve must follow years of clinical experience, in vitro flow analyses give valuable information prior to implantation. Laser Doppler velocimetry and computational fluid dynamics enable quantitative flow analyses to be made in vitro. Whilst these techniques highlight features such as areas of stasis, turbulence and high shear which may predispose to thrombus formation, the complex and time varying nature of the flow through valves makes it difficult to predict accurately potential sites of thrombus deposition and accumulation.

Methods: A technique is described which uses enzyme activated milk as a coagulable blood analogue to indicate flow related clotting. Milk flowing past a test valve or object was activated to clot downstream of the test piece after a certain time period. Milk clot was deposited clot at sites determined by the local flow disturbances. Milk clotting patterns produced on and around standard objects were compared with the transient flow patterns predicted around identical configurations to test the validity of computational flow analyses for predicting flow disturbances leading to clotting. Milk clots on valves were compared with examples of thrombus found on explanted valves of the same design.

Results: The sites of deposition were consistent with the predicted flow patterns around the two configurations of flow obstruction studied. Milk clotting patterns on valves corresponded with the early stages of thrombus on explanted valves of the same design.

Conclusions: Whilst a coagulable milk mixture may be used to evaluate the risk of flow induced clot adherence, care must be taken when extrapolating to the clinical situation as other factors such as material properties, blood chemistry and concomitant disease must also be considered.

How to cite: Keggen, L. A., Black, M. M., Lawford, P. V., Hose, D. R., & Strachan, J. R. (1996). The use of enzyme activated milk for in vitro simulation of prosthetic valve thrombosis. The Journal of heart valve disease, 5(1), 74–83.

Background and aim of the study: The non-invasive, in-vivo assessment of prosthetic valve function is compromised by the lack of accurate measurements of the transvalvular flow fields or hemodynamics by current techniques. Short echo time magnetic resonance imaging (MRI) may provide a method for the non-invasive, in vivo assessment of prosthetic valve function by accurately measuring changes in the transvalvular flow fields associated with normal and dysfunctional prosthetic valves. The objectives of these in vitro experiments were to investigate the potential for using MRI as a tool to measure the complex flow fields distal to replacement heart valves, and to assess the accuracy of MRI velocity measurements by comparison with Laser Doppler Anemometry (LDA), a gold standard.

Methods: The velocity fields downstream of tilting disc, bileaflet, ball and cage, and pericardial tissue valves were measured using both three-component LDA and MRI phase velocity encoding under a steady flow rate of 22.8 l/min, simulating peak systolic flow. The valves were tested under normal and stenotic conditions to assess the MRI capabilities under a wide range of local flow conditions, velocities and turbulence levels. A new short echo time MRI technique (FAcE), which allowed velocity measurements in stenotic jets with high turbulence, was tested.

Results: Good overall agreement was obtained between the MRI velocity measurements and the LDA data. The MRI velocity measurements adequately reproduced the spatial structure of the flow fields. In most cases peak velocities were accurately measured to within 15%.

Conclusions: The results indicate that the FAcE MRI method has the potential to be used as a diagnostic tool to assess prosthetic valve function.

How to cite: Fontaine, A. A., Heinrich, R. S., Walker, P. G., Pedersen, E. M., Scheidegger, M. B., Boesiger, P., Walton, S. P., & Yoganathan, A. P. (1996). Comparison of magnetic resonance imaging and Laser Doppler Anemometry velocity measurements downstream of replacement heart valves: implications for in vivo assessment of prosthetic valve function. The Journal of heart valve disease, 5(1), 66–73.

Background and aims of the study: The main disadvantage of xenograft bioprosthetic heart valves is limited durability due to tissue degeneration.

Methods: To study possible alternatives to conventional bioprosthetic heart valves, glutaraldehyde preserved porcine aortic valve leaflets were precoated with fibronectin-heparin and acidic fibroblast growth factor (aFGF) to enhance cell proliferation. Furthermore, different methods of storage and preservation (1.0% benzoic acid, 1.0% sorbic acid, 0.05% and 0.5% dialdehyde starch) were compared to conventional preservation procedures. Valve leaflets were lined with adult human saphenous vein endothelial cells (1.0 x 10(4) AHSVEC/cm2). The different methods of preservation were compared to leaflets precoated additionally with 1.0 microgram/cm2 aFGF protein, derived from gene manipulated bacteria. Endothelial growth kinetics were studied and verified by an increase of activation following 3H-thymidine incorporation, while the maintained metabolic cell activity was demonstrated by prostacyclin (PGI2) release measurements. In subsequent experiments in vivo degeneration and mineralization were determined by subcutaneous implantation in rats for up to 40 days.

Results: Our results favored alternatively treated valve leaflets; endothelial cells grew to persistent monolayers in six to 12 days in contrast to conventional processed leaflets where no endothelial cell growth was possible. Atomic absorption spectroscopic assessment of subdermal valve implants showed significantly lower contents of calcium, magnesium and phosphate in alternatively treated leaflets. Furthermore, precoating of alternatively preserved valve leaflets with aFGF protein and endothelial cells improved in vitro and in vivo results significantly.

Conclusions: Our study shows that endothelial cell growth as well as significantly reduced in vivo degeneration and mineralization of valve leaflets may be feasible if bioprosthetic heart valves are processed according to alternative, non-toxic conservation procedures and are precoated with angiogenic growth factor protein.

How to cite: Fischlein, T., & Fasol, R. (1996). In vitro endothelialization of bioprosthetic heart valves. The Journal of heart valve disease, 5(1), 58–65.

Background and aim of the study: The mechanism of valvular bioprostheses calcification is still unknown, but early studies showed increased Gla-protein content in calcified valves. Using an experimental model, which reproduces the clinical process, we therefore analyzed the role of minerals and proteins in bioprosthetic valvular calcification.

Methods: Glutaraldehyde pretreated porcine valves were studied before and after implantation in rats by X-ray, calcium (Ca) and phosphorus (P) measurement, Fourier Transform Infrared (FTIR) spectroscopy, SDS-PAGE and 45Ca ligand blotting of the extracted proteins.

Results: Before implantation, there was no X-ray calcification with very little Ca and P content. After implantation, X-ray calcifications appeared on day seven with increased Ca and P up to day 35 (p < 0.05, ANOVA). FTIR revealed structural proteins alone before implantation, plus minor proportions of lipids on day two, which always preceded Ca and P appearance. Ca and P increased up to day 35, first as amorphous and changed in carbapatite over time. SDS-PAGE before implantation revealed two proteins (66-kD and 54-kD) alone, which were sustained up to day 35. The 66-kD had 45Ca affinity. On day two, many other proteins appeared on SDS-PAGE, four of which (52, 45, 14 and below 14-kD) with 45Ca affinity. Protein pattern did not change from day two to 35.

Conclusions: Valvular bioprosthesis calcification is associated with progressive increase in Ca and P content and at least five calcium-binding proteins: one intrinsic valvular protein, pre-existing to implantation, plus four other, extrinsic valvular proteins adsorbed within the tissue after implantation.

How to cite: Shen, M., Farge, D., Daudon, M., Carpentier, S., Pellerin, M., Lacour, B., Chen, L., Martinet, B., & Carpentier, A. (1996). Proteins and bioprosthetic calcification in the rat model. The Journal of heart valve disease, 5(1), 50–57.

Background and aim of the study: This report reviews our experience with single aortic and mitral valve replacement using the Carpentier-Edwards pericardial valve in a population younger than those of other reports.

Methods: Between 1984 and 1993, 106 patients underwent single aortic or mitral valve replacement using a Carpentier-Edwards pericardial valve. There were 49 aortic and 57 mitral implants. The mean age was 56.7 +/- 14.3 years. The follow up was 100% complete; total follow up being 273.3 patient-years in the aortic (AVR) and 348.0 patient-years in the mitral valve replacement (MVR) group, with a mean of 5.6 and 6.1 years, respectively.

Results: The rates of thromboembolic complications for AVR and MVR were 0.37% and 0.58% per patient year respectively. There were no cases of non-structural dysfunction or anticoagulant-related hemorrhage. The incidences of prosthetic valve endocarditis for AVR and MVR were 0% and 0.86% per patient year respectively. Structural deterioration (calcification and sclerosis) occurred in one patient after AVR and in six patients after MVR, all younger than 60 years of age. Freedom from structural deterioration at five years and 10 years was 100% and 92.9% +/- 6.9% for AVR, and 100% and 76.9% +/- 8.6% for MVR.

Conclusions: In conclusion, the Carpentier-Edwards aortic valve has a low incidence of valve-related complications at 10 years. Although late results of this valve in the mitral position were excellent in the elderly, leaflet degeneration is still a problem in patients under the age of 60 years.

How to cite: Murakami, T., Eishi, K., Nakano, S., Kobayashi, J., Sasako, Y., Isobe, F., Kosakai, Y., Kito, Y., & Kawashima, Y. (1996). Aortic and mitral valve replacement with the Carpentier-Edwards pericardial bioprosthesis: 10-year results. The Journal of heart valve disease, 5(1), 45–49.

Background and aim of the study: The Carpentier-Edwards supra-annular valve was introduced in 1982 and was designed to incorporate the perceived benefits of low pressure fixation and supra-annular configuration. We report an eleven year experience with the valve, concentrating on its use for single valve replacement in the aortic position.

Methods: The prosthesis was implanted in 395 patients at 399 operations between January 1984 and October 1993. Myocardial revascularization was additionally performed in 122 (31%) patients. The mean age of the patient population was 65.4 years (range 22-84 years). Sex distribution was predominantly male (62%). Three hundred and seventy-two patients left hospital (30 days operative mortality 5.8%). The mean follow up extended to 3.4 years per patient, and the cumulative follow up for the series was 1264.3 patient years.

Results: Overall actuarial survival at six years was 75.8% +/- 4.1%. The incidence of valve-related complications expressed in terms of actuarial freedom from complication at six years resulted in a freedom from reoperation of 95.1% +/- 1.8%, from prosthetic endocarditis of 97.2% +/- 1.0%, from thromboembolism and anticoagulant-related hemorrhage of 95.9% +/- 1.3% and 95.8% +/- 1.6%, respectively. Freedom from periprosthetic leak was 99.1% +/- 0.7%, whilst freedom from structural valve deterioration was 97.6% +/- 1.5%.

Conclusions: We conclude that the Carpentier-Edwards supra-annular valve compares favorably with other similar bioprostheses and that changes in its design have not impaired its structural integrity.

How to cite: Wilson, E. S., & Jamieson, M. P. (1996). Carpentier-Edwards supra-annular bioprosthesis in the aortic position. Has altered design affected performance?. The Journal of heart valve disease, 5(1), 40–44.

Background and aim of the study: The mechanism of structural failure of bioprosthetic valves is still not clearly understood. This study was undertaken to assess pure leaflet tear as a mode of failure in porcine and pericardial bioprostheses.

Methods: Of 246 bioprosthetic valves (109 porcine, 137 pericardial) implanted between 1975 and 1991, 101 had to be explanted and served as the study population.

Results: The reasons for valve failure were calcific degeneration in 73, pure tear in 12, and endocarditis in 10. Six other patients had a perivalvar leak. The mean age at operation was 32 years. Freedom from degeneration at 10 years was 45 +/- 7% and from pure tear it was 92 +/- 2%. The hazard functions were strikingly different, as that for degeneration showed a progressive increase while that for pure tear peaked at six years post-implant. The mean age of the patients with pure tear was 41 years and for degeneration it was 24 years (p = 0.00001). The reasons for the difference in hazard function are discussed. The characteristic clinical features of pure tear allow clinical diagnosis in the majority of patients.

Conclusion: Pure tear is the result of uneven tissue loading with tearing occurring at sites of maximal stress. The four possible mechanisms in pericardial valves are (a) intense stress concentration at the top of the stent post (commissure); (b) compression stress below the top of the post; (c) abrasion stress in tissue mounted outside the frame; and (d) increased bending stresses on leaflet opening. In stent-mounted porcine valves, pure tear is related to incorrect mounting or to increased bending stresses.

How to cite: Haziza, F., Papouin, G., Barratt-Boyes, B., Christie, G., & Whitlock, R. (1996). Tears in bioprosthetic heart valve leaflets without calcific degeneration. The Journal of heart valve disease, 5(1), 35–39.

Background and aims: Syncope is a serious complication of aortic stenosis. The aim of this study was to determine whether hemodynamic parameters are useful for estimating the risk of syncope in aortic stenosis.

Methods: In 43 patients with aortic stenosis, cardiac catheterization and echocardiography were performed to measure the pressure gradient across the aortic valve, the aortic valve area, left ventricular mass index, systolic left ventricular wall stress and peak systolic coronary artery flow velocities. Hemodynamic parameters were correlated with syncope and the accuracy of those parameters for determining the risk of syncope were assessed.

Results: Ten out of 43 patients experienced syncope. The highest correlation with syncope was found for systolic left ventricular wall stress (R = 0.74, p < 0.001). In descending order of correlation were peak systolic coronary artery flow velocity (R = 0.68, p = 0.002), the pressure gradient across the aortic valve (R = 0.62, p = 0.01) and the aortic valve area (R = 0.43, p = 0.03). Left ventricular mass index was not significantly correlated with syncope. Multivariate analysis revealed systolic left ventricular wall stress to be the only factor contributing to determining syncope (F-to-remove: 47.8). A discriminative value of > 225 dyn/cm-2 x 103 for left ventricular wall stress identified patients with a history of syncope with a specificity of 97% and a sensitivity of 90%.

Conclusions: Syncope in aortic stenosis is closely correlated to increased left ventricular wall stress and attenuated, peak systolic coronary flow velocities. Cut off values may be used to identify patients with an increased risk of syncope.

How to cite: Omran, H., Fehske, W., Rabahieh, R., Hagendorff, A., Pizzulli, L., Zirbes, M., & Lüderitz, B. (1996). Valvular aortic stenosis: risk of syncope. The Journal of heart valve disease, 5(1), 31–34.

Background and aim of the study: The study analyzes the possible predictive power of different clinical and hemodynamic parameters with regard to hospital mortality after the first tricuspid valve replacement.

Methods: A retrospective study of 62 consecutive patients undergoing tricuspid valve replacement from 1974 to 1994 (7.2% of all tricuspid surgery performed at our institution in this period) was completed. Hospital mortality was 37%. Twenty-three patients died in hospital after the first tricuspid valve replacement (group I) and 39 patients (group II) survived this procedure. Tricuspid insufficiency was organic in 87% of group I and 84.6% of group II.

Results: The parameters showing significant differences between the two groups were NYHA class IV (p = 0.05), severe congestive heart failure (p = 0.02), mean right atrial pressure (p = 0.05), pulmonary arterial resistance (p = 0.006) and mean pulmonary arterial pressure (p = 0.0001). Cardiopulmonary bypass time (p = 0.005) and aortic cross-clamp time (p = 0.05) were longer in group I. Multiple regression analysis showed that the variables with greatest predictive power for hospital death were preoperative functional class, congestive heart failure and mean pulmonary artery pressure. The model gave a p < 0.001, with r = 0.58.

Conclusions: The high hospital mortality rate after tricuspid valve replacement seems to be related to clinical functional class, severe congestive heart failure, and to chronic hemodynamic changes in the right atrium and pulmonary circulation.

How to cite: Poveda, J. J., Bernal, J. M., Matorras, P., Hernando, J. P., Oliva, M. J., Ochoteco, A., & Berrazueta, J. R. (1996). Tricuspid valve replacement in rheumatic disease: preoperative predictors of hospital mortality. The Journal of heart valve disease, 5(1), 26–30.

Background and aims of the study: Earlier surgical intervention to the mitral and/or aortic valve means that it will be uncommon to replace the tricuspid valve (TVR) in a patient who has not had prior open heart surgery.

Methods: We report the short and medium term results of a consecutive series of 14 patients who underwent bioprosthetic TVR between December 1985 and February 1993 at the Hammersmith Hospital, UK. All patients had undergone previous open heart surgery on at least one occasion. Mean patient age was 59 years (range: 45-77 years), 11 were female and three were male. Ten patients (72%) were in New York Heart Association class III or IV preoperatively, eight patients were first time reoperations and six patients were second time reoperations.

Results: Hospital mortality was 50% (7/14). Of these seven patients, six were in NYHA class III or IV preoperatively, and three were second time reoperations. There were no 'on-the-table deaths' and no patients required reoperation for bleeding or permanent pacing. For the patients discharged from hospital, the mean follow up was 46 months (range 9-84 months) and it was 100% complete. There was no significant difference in the preoperative assessment data between the hospital mortality group and the patients who left hospital (p > 0.05). Of the patients discharged, four (57% of this group, 28.5% of all patients) showed an improvement in NYHA classification and all patients reported a reduction in peripheral edema. Amongst the survivors there was 100% freedom from valve related complications. Within this group there have been three deaths since discharge, all due to biventricular failure.

Conclusions: From this study we conclude that TVR in patients who have had prior cardiac surgery is a high risk procedure. Nonetheless, amongst survivors, benefit may be gained by either a reduction in peripheral edema and/or an improvement in NYHA class.

How to cite: Hornick, P., Harris, P. A., & Taylor, K. M. (1996). Tricuspid valve replacement subsequent to previous open heart surgery. The Journal of heart valve disease, 5(1), 20–25.

Background and aim of the study: Although depressed left ventricular ejection fraction is present in a considerable proportion of subjects with pure mitral stenosis (MS), its mechanisms are not clearly identified. The purpose of this study was to identify determinants of depressed ejection fraction in young patients with isolated mitral stenosis in sinus rhythm.

Methods: We retrospectively analyzed 320 records of patients with MS (mitral valve area < or = 2.0 cm2) who underwent invasive diagnostic procedure in our center. Of these 39 subjects aged 20-40 years with isolated MS in sinus rhythm were selected for the final analysis.

Results: An ejection fraction not exceeding 50% was found in 12 patients (group A). When comparing group A to the remainder (group B), group A patients had lower left ventricular end-diastolic volume indices (60.5 +/- 21.6 ml/m2 vs. 76.1 +/- 16.1 ml/m2, p = 0.02) and stroke volume indices (28.0 +/- 10.4 vs. 47.9 +/- 12.0, p < 0.001). No significant differences between the groups in patients' age, end-systolic volume index, mitral valve area, mean transmitral gradient, left-sided cardiac pressures, pulmonary wedge pressure, systemic vascular resistance, and cardiac output were found. Indices of left ventricular isovolumic contraction and relaxation as well as end-systolic indices of left ventricular function were also comparable. Group A had significantly higher pulmonary vascular resistance, pulmonary artery pressures, and higher heart rate. An approximate index of left ventricular compliance was significantly lower in group A with similar left ventricular minimal and end-diastolic pressures.

Conclusions: Depressed ejection fraction in pure mitral stenosis with preserved sinus rhythm seems attributable to left ventricular underfilling that appears to be precipitated by other factors in addition to a narrowed mitral orifice, e.g. decreased passive left ventricular compliance and/or altered interventricular interactions.

How to cite: Surdacki, A., Legutko, J., Turek, P., Dudek, D., Zmudka, K., & Dubiel, J. S. (1996). Determinants of depressed left ventricular ejection fraction in pure mitral stenosis with preserved sinus rhythm. The Journal of heart valve disease, 5(1), 1–9.