- Email: [email protected]
Is zero-pressure fixation of bioprosthetic valves truly stress free?
Is zero-pressure fixation of bioprosthetic valves truly stress free? Zero-pressure fixation has often been referred to as stress-free fixation, which implies that no leaflet stresses exist during the fixation process. The two aortic valve cusp layers, the fibrosa and the ventricularis, however, are believed to produce mutually opposing forces within the valve cusps. Residual stresses may therefore exist even during zero-pressure fixation. We first verified the presence of such internal stresses by separating the layers of pig aortic valve leaflets and measuring dimensional changes. In the 11 specimens examined, the fibrosa expanded radially by 30 % ± 13 % (mean ± standard deviation), whereas the ventricularis contracted by 12% ± 4%. The ventricularis also contracted circumferentially by 13% ± 3%. We measured the extensibility of 120 fresh and glutaraldehyde-fixed fibrosa and ventricularis components to investigate the mechanical effects of glutaraldehyde fixation under such internal stresses. We also tested the layers from leaflets that were fixed whole. We compared the extensibility of the fibrosa and the ventricularis, each fixed in the presence and absence of residual stresses, and found that, in the radial directions, the ventricularis from valve cusps that were fixed whole was less extensible than fresh ventricularis (35.4 % versus 63.7% strain to high-modulus phase, p < 0.00001). The fibrosa from cusps that were fixed whole, however, was more extensible than fresh fibrosa (39.2% versus 29.5 % strain, p < 0.0122 radially; 12.8 % versus 8.2 % strain, p < 0.0001 circumferentially). The ventricularis became less extensible because it was fixed under tension, and the fibrosa became more extensible because it was fixed under compression. This study therefore demonstrates the presence of residual tensile and compressive stresses in the ventricularis and fibrosa, even when the leaflets are relaxed. Zero-pressure fixation cannot therefore be considered truly stress free, in the engineering sense, because residual internal stresses affect collagen fiber crimp and change the extensibility of the fibrosa and the ventricularis. (J THORAC CARDIOVASC SURG 1993;106:288-98)
Ivan Vesely, phD,a, b,c Alan Lozon," and Eric Talman.v" London, Ontario, Canada
Zero-pressure fixation has been introduced as a means of reducing leaflet stresses during the preparation of porcine bioprosthetic valves, This process is often referred to as stress-free fixation. I This may not be an accurate From The John P. Robarts Research Institute" and the Departments of Electrical Engineering" and Medical Biophysics," University of Western Ontario, London, Ontario, Canada. Supported in part by a grant-in-aid from the Heart and Stroke Foundation of Ontario. Dr. Vesely is a Research Scholar ofthe Foundation, and Mr. Lozon was supported in part by a Summer Research Student Scholarship from the Natural Sciences and Engineering Research Council of Canada. Received for publication Dec. 4, 199\. Accepted for publication July 27, 1992. Address for reprints: Ivan Vesely, PhD, The John P. Robarts Research Institute, P.O. Box SOlS, London, Ontario, Canada N6A 5K8. Copyright @ 1993 by Mosby-Year Book, Inc. 0022-5223/93 $\.00 +.10
288
12/1/41330
description of this complex and, as yet, poorly understood process. Clarification of some of the terminology used in valve fixation research would improve understanding of the differences among the many valve fixation processes currently available. The clinical perception of the term stress-free fixation may be that no transmural pressure head is applied to the valve during the glutaraldehyde treatment process. Indeed, low-pressure fixation means that a low-pressure head of 1 to 2 mm Hg is used to close the valve leaflets and therefore stress the leaflet material very little. In the strictest sense, however, stress-free fixation should mean that no external or internal forces are applied to the valve leaflet material. Because the valve cusps are composite structures containing a collagenous fibrosa, elastic ventricularis, and a loose, watery spongiosa.? none of these components would be stressed if zero-pressure fixation were truly stress-free. Altered collagen morphology and leaflet mechanics are
The Journal of Thoracic and Cardiovascular Surgery Volume 106, Number 2
the undesirable consequences of fixing valves under hydrostatic preload, Aortic valve cusps are constructed primarily from crimped collagenfibers/? that can stretch readily when the valve is pressurized with a hydrostatic head as low as several millimeters of mercury.'" ? This is thenatural mechanism that enables the leafletsto stretch, coapt, and seal off the ventricle during early diastole. In the preparation of bioprostheses, the valves are usually fixed closed under a hydrostatic pressure head of fixative. In such a scheme,the collagen fibers becomecrosslinked together with their crimp stretched out and do not recoil when the hydrostatic pressure is removed.Consequently, theleaflets losetheir complianceand respondvery differently from natural, unfixed valves, such as aortic valve allografts," New fixation techniques have therefore been designed to preserve natural collagen fiber crimp geometryand leaflet mechanics by reducing leaflet stress during fixation." Zero-pressure fixationtherefore has implicit advantages over low-pressure fixation. When zeropressure fixation is presented as stress-free fixation, however, a confusion ofterminology results. Ifzero-pressure fixation is truly free of stress, then zero-pressurefixed tissuewould have a collagencrimp and compliance very similar to those of fresh tissue. Althoughsome tensile testing studies have found little difference betweenfresh and zero-pressure-fixedporcine aortic valves," a deeper look into leaflet micromechanics isrequiredto fullyevaluate these newvalveconcepts.I D- 13 We know that the aortic valve contains both elastin and collagen-two proteins that react differently to glutaraldehyde fixation. The leaflets themselves are highly complex and wellorganized and contain three distinct fibrous layers: the fibrosa, the ventricularis, and the spongiosa/ (Fig. I). The fibrosa, a thicker, collagen-dominatedlayer, is believed to resist most of the tensile forces during valve closure. The ventricularis, on the other hand, contains both elastin and collagen- 15 and is highly extensible in the radial direction. In our previous work,13,16 we have shown that many of the wrinkles and folds in the fibrosa are maintained by virtue of its attachment to the ventricularis. When these two layers were separated, the fibrosa spontaneously straightened out and the ventricularis shrank. Some internal forces must have therefore been responsible for holding the fibrosa in its folded configuration. Internal stressessuch as these are not uncommon in biologic tissues. Fung and associates'" 18 have reported residualstressesin arteries and trachea and have tested for these stressesby selectively cutting the material and observing the retraction that takes place. If a similar phenomenon exists in the aortic valve leaflet, then fixing valves at zero pressure cannot guarantee truly stress-free fixation.
Vesely, Lozon, Talman
289
Fig. 1. Cutaway through leaflet and aortic wall showing internal configuration of fibrosa, spongiosa, and ventricularis. Nearthecentral portion ofleaflet, fibrosa exists asa folded sheet that unfolds as the leaflet stretches radially. We have therefore hypothesized that residual tensile and compressive stresses do exist within the fibrosa and the ventricularis of aortic valveleaflets. If this is the case, such stresses will change collagen fiber crimp geometry and the extensibilityof the leaflet layers after fixation. A clearer understanding of aortic valve structure and function at the fibrous level and a better definition of leaflet stressesduring fixationshould enable better evaluation of new valve designs. Methods
Thisstudy consisted of two parts. In the first part, wemeasured therelative amounts ofresidual strain that exist within the components of aortic valve cusps. Thiswas assessed byopticallymeasuring thedimensions of fibrosa andventricularis before andaftertheirseparation. Afterthe presence ofresidual strains was established, wemeasured theeffects oftheirpresence onthe extensibility ofthevalve leaflet components during fixation. This effect was assessed by means of uniaxial tensile tests. Microdissection. Whole hearts were obtained from freshly slaughtered pigs at the abattoir. These pigs ranged in agefrom 3 to 6 months and weighed an average of 100 kg. The hearts were transported to the laboratory, and the aortic valve cusps were cut out and kept in chilled pH-balanced saline solution during handling. The cusps were placed on a plastic dissecting
290
Vesely, Lozon, Talman
The Journal of Thoracic and Cardiovascular Surgery August 1993
Fig. 2. Microdissection process. By gently grasping ventricularis with dissecting forceps and pulling upwards, we exposed connecting fibers between the two layers, which could be severed with microscissors. In most cases, continuous pulling on tissue with instruments enabled layers to be torn apart along spongiosa, apparently the natural cleavage plane. board with the ventricularis side up. By gently grasping and lifting the ventricularis, we were able to expose the fibrous connections that spanned the spongiosa and connected the two layers. By carefully pulling on the ventricularis with blunt forceps, we separated the two layers along their natural cleavage plane, which appeared to be the midline of the spongiosa. Where such separation was not possible, for example, near the edges of the cusp, the fibers that held the layers together were cut one by one with microscissors. Such a process enabled the ventricularis to be separated from the fibrosa with minimal tearing (Fig. 2). Saline solution was applied repeatedly to keep the tissue moist. All dissection was performed with the use of a stereo microscope and, after dissection, the tissue samples were examined at the highest magnification available to ensure that the strips to be tested were not damaged in any way. This technique was slightly modified according to the size of the test specimens, as described later. Part 1. Optical measurement of dimensional changes. If the fibrosa and the ventricularis are under mutual compression and tension, as we have suggested.P: 16 then a change in dimensions should occur when these layers are' separated and the stresses are relieved. We therefore trimmed the whole leaflets into rectangular specimens with the cut edges oriented along the ridges visible on the aortic surface. The axes of the specimens were oriented in the standard radial and circumferential directions, and the specimens were then transferred to glass slides by floating them in saline solution and allowing them to settle onto the slides. In this way, the tissues were handled for a short time, and their natural dimensions were affected as little as possible. The slides were placed under a dissection microscope and a Sony
CCD video camera (model AVC-D?, Sony of Canada Ltd. Willowdale, Ontario) was connected to the auxiliary image port; an image of the rectangular piece of tissue was acquired by means of an image analysis package (JAVA, Jandel Scientific, Corte Madera, Calif.). These images were saved on disk on a personal computer for later analysis. The rectangular strip was then separated into its two layers as described previously, the fibrosa and ventricularis were floated in saline solution and deposited on slides by means of the method previouslydescribed, and similar images were acquired and saved. A calibrated scale was placed beside each specimen for the determination of the true dimensions regardless of the magnification of the microscope (Fig. 3). After images of all specimens were acquired, their dimensions in the radial and circumferential directions were measured with JAVA software. Because of the limited precision in cutting the specimens perfectly square, the accuracy of the dimension measurements is unlikely to be better than O.5mm. Part 2. Changes in extensibility. Ifthe fibrosa and the ventricularis are prestressed and the collagen fiber crimp is altered, then leaflet components fixed in glutaraldehyde under these conditions should have mechanical characteristics different from those of leaflet components fixed truly stress free. We assessed the effects of residual prestressing on the extensibility of porcine aortic valve leaflet components by examining the fibrosa and ventricularis individually. A total of three groups of tissue were studied. The first group consisted of fibrosa and ventricularis removed from aortic valve cusps that had been fixed at zero pressure. These tissues were therefore exposed to the naturally occurring internal leaflet stresses. The second group
The Journal of Thoracic and Cardiovascular Surgery Volume 106, Number 2
Vesely, Lozon, Talman
29 I
Fig. 3. Sections of valve cusps cut to measure elongation after separation of fibrosa and ventricularis. Whole tissue is shown on left and fibrosa alone on right. Note expansion of fibrosa in radial direction. Scale was imaged with tissue and used for calibration. consisted of fibrosa and ventricularis fixed in glutaraldehyde after they were separated from each other. These tissues therefore had no stresses imposed on them whatsoever. The third group included the same fresh specimens tested immediately after dissection. After testing, these fresh tissues were allowed to relax for 24 hours in a saline bath before fixation in glutaraldehyde. The fresh and separately fixed tissues acted as matched controls for each other. The three groups were termed ( 1) wholeor intact-fixed, (2) glutaraldehyde-treated, and (3) fresh fibrosa and ventricularis. Each of these was tested in the radial and circumferential directions. Ten specimens were in each group, for a total of 120 specimens. The fixative used in all cases was 0.5% glutaraldehyde buffered in phosphate buffer (0.067 moll L) at pH of 7. All tissues were fixed for a minimum of 24 hours before any handling or testing. Tensile testing. The three groups of tissues described previously were cut into 5 mm wide circumferential and 10 mm wide radial strips (Fig. 4). This was done to ensure that the width of the final strips was consistent, regardless of the degree of expansion or contraction that occurred during the dissection process.The extensibility of these strips was then measured with an Instron tensile testing machine (model 1125, Instron Corp., Canton, Mass.). The precision in cutting the strips to width was estimated to be ± 0.2 mm. Thickness was measured by a special device constructed for this purpose, II and the gauge length of the test strip was determined with a calibrated scale attached to the tissue-clamping mechanism. The precision of the thickness measurements and the gauge length were estimated to be ± 0.05 mm and ± 0.25 mm, respectively. The strips of tissue
.-_ •
• : ••
•• : ••
_-_._--------~
Circumferential
Radial
.... _---------_ .....- • •• Fig. 4. Whole leaflet or layers after dissection ~_
• .1
(top), before cutting specimens for tensile testing. Leaflet was trimmed to 5 mm wide circumferential strips and 10 mm wide radial strips, with circumferential and radial directions as shown. Nodulus Arantii was avoided in test strips.
292
The Journal of Thoracic and Cardiovascular Surgery August 1993
Vesely, Lozon, Talman
water level
..
I I....-
_ -
- - rod
_ - clamp
!-"...~ -
clamp
Side view Tissue straight
Tissue wrinkled
Tissue stretched
"0 C'O
o
...J
®
CD Buoyancy slope
Elongation
o.z rnm uncertainty
Fig. 5. Technique used to measure zero-load length of specimens. During first phase, only wrinkles were eliminated from tissue, as in 1 and 2. Load increased only because of decrease in buoyant force on clamp assembly as its parts were withdrawn from water. When tissue was completely straight and offered resistance, curve deviated from straight line as in 3. We then began to measure extensibility of specimens. Choosing gauge length at arbitrary level of load would have underestimated extensibility of specimens. were clamped between stainless steel jaws lined with No. 320 emery cloth to prevent slippage. The tissue was preconditioned by repeated stretching at an elongation rate of 10 mm/rnin to a maximum load of 0.98 N (100 gm) until reproducible load/ elongation curves were attained. The load/elongation curves were transformed to engineering stress versus strain curves by normalizing with the initial, unstrained dimensions of the specimens. All testing was done in physiologic saline solution heated to 37° C. Gauge length measurements. Because many of our conclusions for this study are based on the measured extensibility of the tissue strips, we took particular care in determining the starting length (gauge length) of the specimens at loads as close to zero as possible. The measurement of stress at some arbitrary value, for example, 0.5 gm, would produce consistent results but would prestretch the tissue strips, overestimate the gauge length,
and underestimate the extensibility. For the determination of the zero-load length, the tissue was clamped between the grips of the tensile testing machine, and the grips were moved closer together until the tissue was slightly wrinkled. The gain on the load amplifier was set at the maximum level, and the grips were slowly moved apart (Fig. 5). As the wrinkles in the tissue straightened out, no increase in tension was recorded at first. The only change was a slight increase in load as the tissue grip shaft was withdrawn from the water bath; this resulted from a slight decrease in the buoyant force on the grip shaft. During this phase, the unwrinkling of the tissue did not contribute enough force to affect the linear increase in load. After the tissue straightened and began to generate tensile resistance, the plotted curve sharply deviated from the straight line. This was the point at which we chose to measure the gauge length. The uncertainty in the measurement of gauge length was estimated
The Journal of Thoracic and Cardiovascular Surgery Volume 106, Number 2
Vesely, Lozon, Talman
29 3
o+--------,--~==:::::;======----.------------J
o
20 strain in %
10
30
40
Fig. 6. Method to calculate extensibility of test strip--the point at which a tangent drawn to the stress/strain curve at 300 kPa crossed the x axis, as shown. Extensibility was expressed as % strain. In this example, the extensibility is 31%.
Table I. Mean changes in length of leaflet components after their separation Change in dimensions (%) Tissue Fibrosa Ventricularis
Direction
No.
Mean
SD
Circumferential Radial Circumferential Radial
II II II II
- 4.78 30.47 -13.37 -12.29
5.11 13.19 2.74 3.75
A negative number denotes contraction, and a positive number denotes expansion. All values are change in percentage relative to dimensions when the two layers were connected. SD, Standard deviation.
to be ± 0.1 mm. For the shortest specimen, with a gauge length of only 7 mm, our estimate of gauge length would have an error of only 1.4%. In our view, this technique is the least likely to underestimate the extensibility of highly extensible tissue, especially when compared with other techniques that measure gauge length at an arbitrary load. This technique is therefore particularly good for analyzing the initial, high-extensibility region of the stress/strain curve of radial strips of aortic valve cusp tissue. We admit, however, that our technique can overestimate tissue extensibility if the tissue is not positioned squarely between the tissue grips. After much practice, however, this effect was kept to a minimum. This protocol has been used in our laboratory for several years and has produced consistent results. The collagen fiber crimp was assumed to be proportional to the extensibility of the tissue strips, which was measured from
Fig. 7. Inward contraction of specimens immediately after dissection caused by tension in ventricularis released by cut. Contraction produced slight bowing of specimen and skewing of cut face inward.
the length of the low-modulus phase of each stress/strain curve. This was obtained from the intersection of the x axis with a line drawn tangent to the curve at a stress of 300 kPa (Fig. 6). The extensibility was expressed as percentage of strain. A change in
The Journal of Thoracic and
294
Vesely, Lozon, Talman
Cardiovascular Surgery August 1993
Mean and SEM Circumferential Fibrosa
Radial Fibrosa Circumferential Ventricularis
Radial Ventricularis
o -20
-10
o
10
20
30
% change in length
•
Circumferential Fibrosa
•
10
20
30
Radial Fibrosa
~ Circumferential Ventricularis ~ Radial Ventricularis
Fig. 8. Mean changes in length of fibrosal and ventricularis leafletcomponents after separation.Error bars denotestandard error of mean. Slight contraction of fibrosa in circumferential directionwas unexpected and likelyresulted from realignment of fibers rather than major contractionof collagen fibercrimp. extensibility would indicate to what extent the collagen crimp wasaffectedby the fixation process. Becauseall specimens were analyzed with the same method,our resultsare internallyconsistent, and comparisons between groups are valid. Results General observations. The prestressing of the fibrosa and the ventricularis became apparent as soon as rectangular specimens were cut from the leaflets and observed at high magnification. The cut edges of the leaflets were not square but were skewed toward the ventricularis, as shown in Fig. 7. This occurred when a segment of tissue was cut and separated from surrounding tissue. Because no loads could exist at the edges of the tissue, the edges moved laterally in response to the stress field in the interior of the specimen. The ventricularis therefore contracted slightly, and the fibrosa expanded. For consistency, the dimensions of the fibrosa side were used as the dimensions for the whole tissue. Part 1. Dimension changes. The mean dimensions of the rectangular specimens before dissection were 8.37 ± 1.91 mm in the circumferential direction and 6.10 ± 1.53 mm (mean ± standard deviation) in the radial direction. After separation of the two layers, the fibrosa expanded considerably (by 30%) in the radial direction. This was statistically significant (p < 0.00001). In the circumferential direction, however, a small but statistically significant shrinkage (less than 5%) was observed after dissection (p < 0.02; Fig. 8, Table I). The ventricularis, on the other hand, contracted radially by
40
50
60
70
Strain in %
40
o
Fresh
•
Treated
•
Whole
Fig. 9. Mean extensibility of fibrosa and ventricularis, tested in radial and circumferential directions in freshand glutaraldehyde-treatedstates. Treated denoteslayersthat havebeenfixed whileseparated, and whole denotes that they have been fixed whilestill connected together. Statistical comparisons between multiplegroups are shownin Table III. Note increased extensibility of fibrosa from whole-fixed leaflets in both radial and circumferential directions, compared with both fresh and separately-treated specimens. Error bars denotestandard error of mean. 12% (p < 0.00001) and circumferentially by 13% (p < 0.00001). Because the leaflet components spontaneously changed their length after separation from each other, the fibrosa and the ventricularis must be naturally preloaded; most notably, the fibrosa was preloaded in the radial direction. This preload can be considered residual stress that may affect tissue behavior after fixation. Part 2. Tensile testing. The effects of prestressed and stress-free fixation on the extensibility of the leaflet components were evaluated from the changes in the measured extensibility of rectangular test strips. The radial extensibility of the fibrosa did not change when it was fixed separately, but increased significantly when it was fixed as part of the whole leaflet (39% whole-fixed versus 29% fresh;p < 0.0122) (Fig. 9, Tables II, III). A similar effect occurred in the circumferential direction. The separated fibrosa, fresh and treated leaflets, had similar extensibilities (8% and 7%, respectively,) but fibrosas from leaflets fixed whole were more extensible (13%, P < 0.0003 or better). An opposite effect occurred in the radial ventricularis. Ventricularis from the whole-fixed leaflets had a significantly decreased extensibility when compared with fresh leaflets (35% whole-fixed versus 64% fresh; p < 0.00001). There was no statistically significant difference in the extensibility of any ventricularis specimens in the circumferential direction.
The Journal of Thoracic and Cardiovascular Surgery Volume 106, Number 2
Vesely, Lozon, Talman
295
Table II. Extensibilities offibrosa and ventricularis Extensibility (%) State Fresh
Layer Fibrosa Ventricularis
Treated
Fibrosa Ventricularis
Whole-fixed
Fibrosa Ventricularis
Direction
No.
Mean
SD
Circumferential Radial Circumferential Radial Circumferential Radial Circumferential Radial Circumferential Radial Circumferential Radial
10 10 10 10 10 10 10 10 II 9 10 10
8.150 29.500 13.460 63.660 7.140 33.150 19.420 57.810 12.750 39.230 14.650 35.410
2.533 9.417 5.450 11.260 2.916 8.243 6.964 9.096 2.337 5.036 5.025 6.461
Treated, Layers that were fixed while separated; Whole fixed, layers that were fixed while connected. See also Fig. 9. SD, Standard deviation.
In summary, when leaflet layers were separated before glutaraldehyde fixation, no significant changes in extensibility were observed, but leaflet layers fixed together did show changes. This occurred both radially and circumferentially in the fibrosa and radially in the ventricularis. Apparently, the presence of residual stresses within aortic valve leaflets during zero-pressure fixation significantly changed the extensibilities of both the fibrosa and the ventricularis components. Discussion It has been shown that reducing fixation pressures lowers leaflet stresses, minimizes the distortion of the collagen fiber crimp.> 7 and preserves a significant portion of the natural mechanical function of aortic valve cusps. The term zero-pressure fixation, however, has sometimes been linked with the term stress-free fixation. Such a linkage suggests that zero-pressure-fixed valves have unchanged collagen fiber crimp structure and therefore distensibility and compliance characteristics similar to those of fresh tissue. This has been demonstrated to some extent with standard biaxial tensile testing." Recent work in basic biomechanics, however, suggests that most biologic structures, including the aortic valve, are prestressed even in the relaxed state. 13, 16, 18 In this study, we have demonstrated the existence of residual stresses in the fibrosa and the ventricularis and have shown that even zero-pressure fixation does not guarantee truly stress-free fixation. Certain assumptions were made in this study. In part 1, we assumed that the dimensions of the leaflet layers were not altered significantly during handling and that the differences in size observed between the layers after
Table III. Statistical comparisons among the groups shown in Table II Groupings
Comparisons
Extensibility
Circumferential p < 0.00001 vs radial Ventricularis Circumferential p
Fibrosa
NS, Not significant. Level of significance was determined to be 0.05 by means of a small sample t test. Paired t tests were used to compare fresh and treated groups because they were the same tissue.
29 6
Vesely, Lozon, Talman
dissection were not artifactual. We believe that if great care is taken in floating the leaflets onto glass slides, the only force present that could distort the dimensions of the leaflets would be surface tension, and we believe this distortion to be minimal. In part 2, we assumed that the changes in extensibility measured with uniaxial tensile tests were real and that similar changes would exist in the whole aortic valve leaflets when stretched biaxially during normal cardiac function. We should point out, however, that uniaxial tensile tests tend to overestimate tissue extensibility. This results from fiber reorientation as the strip is stretched in one direction only. When stretched biaxially, these fibers reorient much less and offer resistance to stretching at lower extensions. The extensibility values presented in this article were therefore used only as a comparison between groups and do not represent the true extensibility that would be expected from a whole leaflet. We chose uniaxial over biaxial testing because the technique for determining gauge length and measuring the very high extensibility region of the stress/strain curve cannot be readily applied to biaxial testing machines. We also assumed that the measured extensibility of tissue is determined largely by the state of the collagen crimp. In reality, extensibility is only an estimate of collagen crimp because some of the extensibility occurs by means of a realignment of collagen fibers and not as a result of uncrimping. Finally, we assumed that glutaraldehyde fixation crosslinked the collagen fibers in the configuration that existed during the treatment process. If the fibers were uncrimped during fixation, they would have stayed uncrimped, even after the load applied to them during fixation was removed. This enabled us to measure differences inextensibility before and after fixation and to relate these differences to the loads that were present during fixation. Although this assumption is valid for the fibrosa, which does not appear to have much elastin, it may not be entirely valid for the ventricularis. Elastin is immune to cross-linking with glutaraldehyde'? and would therefore attempt to recoil to its original configuration after the load had been removed. Ventricularis fixed in the whole leaflet might recoil and contract after being separated from the fibrosa. Although we did not specifically test for this, we did not observe any recoil during dissection of the wholefixed cusps, as we did with the fresh cusps. It is therefore possible that the extensibility of the ventricularis from leaflets fixed whole may be overestimated because of this phenomenon. Nevertheless, because we have shown that whole-fixed ventricularis has less extensibility than does fresh or fixed ventricularis, the possible recoil of the whole-fixed ventricularis does not alter our conclusions.
The Journal of Tboracic and Cardiovascular Surgery August 1993
In the context of the previous assumptions, our results provide additional insight into the internal micromechanics of aortic valve leaflets and the effects of conventional glutaraldehyde-fixation techniques. The failure of existing bioprosthetic valves to provide satisfactory service and durability may be related to our limited knowledge of leaflet micromechanics and the glutaraldehyde-fixation process itself. Some of the pioneering work by Broom and associates'v? has demonstrated that valve tissues fixed under high pressures do not have natural mechanical function and may therefore wear out prematurely. Our earlier studies 10, II have suggested that the failure of glutaraldehyde-fixed tissue may occur as a result of flexural damage to the fibrous components of the leaflet material. Unfixed materials, when bent, do not exhibit the sharp kinking and compressive buckling of fixed tissues. 10, 20 If mechanical damage brought about by unnatural mechanical characteristics does indeed contribute to bioprosthetic valve failure, then such mechanical damage can be minimized by means of properly controlled glutaraldehyde-fixation techniques. Understanding the function of the valve leaflets at the fibrous level is therefore a prerequisite to developing a fixation process that can preserve the natural structure and function of the valve leaflet material. We now know that even internal leaflet stresses during fixation affect valve function at the micromechanicallevel. Because the fibrosa taken from a whole-fixed cusp is more extensible than one that is fixed alone, the fibrosa must have been under radial compression by virtue of its attachment to the ventricularis. The "freezing" of these fibers during fixation under compression increased the amount of radial fiber crimp and therefore increased their radial extensibility. In the ventricularis, this phenomenon occurred in the opposite direction. Because the ventricularis from whole-fixed cusps had a lower extensibility than either the fresh or fixed tissue, the ventricularis must have been under tension during fixation. Tension reduces the collagen fiber crimp and the resultant extensibility. This residual tension in ventricularis may also explain why the leaflet has a tendency to flip and curl toward the ventricularis after it is cut free from the aortic root. In most cases, the changes in extensibility measured with uniaxial tensile testing reflected the preload, which could be inferred from the changes in dimensions after dissection. The fibrosa was under compression in the radial direction and was naturally expanded when released from the ventricularis; it had an increased collagen fiber crimp when fixed while part of the whole leaflet and therefore had an increased extensibility. The ventricularis, on the other hand, was under tension in the radial
The Journal of Thoracic and Cardiovascular Surgery Volume 106, Number 2
direction, contracted when released from the fibrosa, had a decreasedcollagen fiber crimp when fixedas part of the whole leaflet,and therefore had a decreased extensibility. A similar scenario occurred for the ventricularis in the circumferential direction but not for the fibrosa. The circumferential fibrosa was expected to expand circumferentiallywhen dissected free but, instead, shrank slightly. Although the shrinkage was less than 5%, it was statistically significant. This finding remains unexplained but may be related to the interactions between collagen fiber crimp and fiber migration during tensile testing, as mentioned previously. These interactions are still not well understood. What does zero-pressure fixation really mean? In this study, we have concluded that zero-pressure fixation does not ensure truly stress-free-fixed valves. Although there may not be any external forces applied to the valve cuspsduring fixation, internal forces are present by virtue of the interactions between the fibrosa and the ventricularis. When the whole valve is fixed, even under zero transmural pressure, the ventricularis is fixed under tension and the fibrosa is fixed under compression. Because the crimp of the collagen fibers is "frozen" into the tissue during glutaraldehyde fixation, zero-pressure-fixed valves cannot have the same mechanical behavior as the fresh valves. Although the wide-ranging effects of this typeof fixationare still not wellunderstood, zero-pressure fixation stillaffects leaflet mechanics much less than lowpressure fixation.P Further confusion may arise from reference to leaflet pressure and transmural pressure. Recently,the importance of aortic root geometry has been recognized." and some valvesare being fixed at high root pressures to simulate diastolic root geometry while still maintaining pressures across the leaflets that are very close to zero. Such valves can be correctly called zeropressure-fixedvalvesif the differencesbetween the stresses within the leaflets and the aortic wall are acknowledged. New fixation schemes. Because it has been shownthat stresses during fixation reduce collagen fiber crimp and produce leaflets with unnatural mechanical function.v? the control of fixation pressures has been an important focusof many heart valve manufacturers. More recently, however, it has been suggested that other leaflet micromechanics,in addition to collagen fiber crimp, are important in defining physiologic valve function. 10, II, 14,22 The interaction of the fibrosa and the ventricularis is one such critical issue.P: 16 More innovative fixation schemes may have to be applied to preserve the natural micromechanical function of aortic valve cusps. Dynamic fixation, a technique first proposed by US,22 has been shown to
Vesely, Lozon, Talman
297
improve some of the mechanical characteristics of aortic valve tissue.v' Perhaps the best type of fixation is none at all. In the future, it may be more appropriate to attempt to eliminate the problem of antigenicity of xenograft material by means of techniques other than glutaraldehyde fixation and to use a more direct approach, such as antigen extraction and masking.i" We are grateful for the assistance of Dr. D. R. Boughner in the composition of this manuscript. REFERENCES I. JaffeWM, Barratt-Boyes BG,SadriA, Gavin JB, Clover-
dale HA, NeutzeJM. Earlyfollow-up of patients with the Medtronic Intact porcine valve. J THoRAc CARDIOVASC SURG 1989;98: 181-92. 2. Gross L, Kugel MA. Topographic anatomy and histology ofthe valves in the humanheart.Am J Pathol1931 ;7:44556. 3. Clark RE, Finke EH. Scanning and light microscopy of human aortic leaflets in stressed and relaxed states. J THORAC CARDIOVASC SURG 1974;67:792-804. 4. HilbertSL, Ferrans VJ, Swanson WM. Optical methods forthenondestructive evaluation ofcollagen morphology in bioprosthetic heart valves. J Biomed Mater Res 1986; 20:1411-21. 5. Hilbert SL, Barrick MK, Ferrans VJ. Porcine aorticvalve bioprostheses: a morphologic comparison of the effects of fixation pressure. J Biomed Mater Res 1990;24:773-87. 6. Broom ND, Marra D. Effects of glutaraldehyde fixation andvalve constraint conditions on porcine aorticvalve leaflet coaption. Thorax 1982;37:620-6. 7. Broom ND, Thomson FJ. Influence of fixation conditions on the performance ofglutaraldehyde-treated porcine aortic valves: towards a more scientific basis. Thorax 1979; 34:166-76. 8. Vesely I, Gonzalez-Lavin L, Graf D, Boughner DR. Mechanical testing ofcryopreserved aorticallografts: comparison with xenografts and fresh tissue. J THORAC CARDIOVASC SURG 1990;99:119-23. 9. Mayne ASD, Christie GW, Smaill BH, Hunter PJ, Barratt-Boyes BG. Anassessment ofthe mechanical properties ofleaflets from foursecond-generation porcine bioprostheseswith biaxial testing techniques. J THORAC CARDIOVASC SURG 1989;98: 170-80. 10. Vesely I, Boughner DR, Song T. Tissue buckling as a mechanism ofbioprosthetic valve failure. AnnThoracSurg 1988;46:302-8. II. Vesely I, Boughner DR. Analysis of the bending behaviour of porcine xenograft leaflets and of natural aortic valve material: bending stiffness, neutralaxisandshearmeasurements. J Biomech 1989;22:655-71. 12. Vesely I. Analysis of the Medtronic Intact bioprosthetic valve: effects of "zero-pressure" fixation. J THORAC CARDIOVASC SURG 1991;101:90-9.
298
Vesely, Lozon, Talman
13. Vesely I, Noseworthy R. Micromechanics of the fibrosa and ventricularis of aortic valve leaflets. J Biomech 1992; 25:101-13. 14. Vesely I, Krucinski S, Campbell G. Micromechanics and mathematical modelling: an inside look at bioprosthetic valve function. J Cardiac Surg 1992;7:85-95. 15. Jonas RA, Ziemer G, Britton L, Armiger LC. Cryopreserved and fresh antibiotic-sterilized valved aortic homograft conduits in a long-term sheep model. J THoRAc CARDIOVASC SURG 1988;96:746-55. 16. Vesely I, Lozon A. Natural preload of aortic valve leaflet components during glutaraldehyde fixation: effects on tissue mechanics. J Biomech 1993;26:121-31. 17. Fung YC, Liu SQ. Change of residual strains due to hypertrophy caused by aortic constriction. Circ Res 1989; 65:1340-9. 18. Han HC, Fung YC, Residual strains in porcine and canine trachea. J Biomech 1991;24:307-15.
The Journal of Thoracic and Cardiovascular Surgery August 1993
19. Fung YC, Sobin SS. The retained elasticity of elastin under fixation agents. J Biomech Eng 1981;103:121-2. 20. Broom ND. An in vitro study of mechanical fatigue in glutaraldehyde-treated porcine aortic valve tissue. Biomaterials 1980;1:3-8. 21. Vesely I, Menkis AH, Rutt B, Campbell G. Aortic valve/ root interactions in porcine hearts: implications for bioprosthetic valve sizing. J Cardiac Surg 1991;6:482-9. 22. Song T, Vesely I, Boughner DR. Effects of dynamic fixation on the shear behavior of porcine xenograft valves. Biomaterials 1990;11:191-6. 23. Mavrilas D, Missirlis Y. An approach to the optimization of preparation of bioprosthetic heart valves. J Biomech 1991; 24:331-9. 24. Vesely I, Noseworthy R, Wilson G. Development of a hybrid xenograft/autograft aortic valve bioprosthesis. Can J CardioI1991;7[suppl]83A.
Bound volumes available to subscribers Boundvolumesof THEJOURNAL Of THORACIC AND CARDIOVASCULAR SURGERY are availableto subscribers(only) for the 1993issues from the Publisher, at a cost of $75.00 for domestic,$98.25 for Canadian, and $93.00for international subscribersfor Vol. 105(JanuaryJune) and Vol. 106 (July-December). Shipping charges are included. Each bound volume contains a subject and author index and all advertising is removed. Copiesare shipped within 60 days after publicationof the last issueof the volume.The bindingis durable buckram with the JOURNAL name, volume number, and year stamped in gold on the spine. Paymentmust accompany all orders. Contact Mosby, SubscriptionServices,11830WestiineIndustrial Drive,St. Louis,Missouri63146-3318,USA; phone I (800) 453-4351 or (314) 453-4351. Subscriptions must be in force to qualify. Boundvolumes are not availablein place of a regular JOURNAL subscription.
Ivan Vesely, phD,a, b,c Alan Lozon," and Eric Talman.v" London, Ontario, Canada
Zero-pressure fixation has been introduced as a means of reducing leaflet stresses during the preparation of porcine bioprosthetic valves, This process is often referred to as stress-free fixation. I This may not be an accurate From The John P. Robarts Research Institute" and the Departments of Electrical Engineering" and Medical Biophysics," University of Western Ontario, London, Ontario, Canada. Supported in part by a grant-in-aid from the Heart and Stroke Foundation of Ontario. Dr. Vesely is a Research Scholar ofthe Foundation, and Mr. Lozon was supported in part by a Summer Research Student Scholarship from the Natural Sciences and Engineering Research Council of Canada. Received for publication Dec. 4, 199\. Accepted for publication July 27, 1992. Address for reprints: Ivan Vesely, PhD, The John P. Robarts Research Institute, P.O. Box SOlS, London, Ontario, Canada N6A 5K8. Copyright @ 1993 by Mosby-Year Book, Inc. 0022-5223/93 $\.00 +.10
288
12/1/41330
description of this complex and, as yet, poorly understood process. Clarification of some of the terminology used in valve fixation research would improve understanding of the differences among the many valve fixation processes currently available. The clinical perception of the term stress-free fixation may be that no transmural pressure head is applied to the valve during the glutaraldehyde treatment process. Indeed, low-pressure fixation means that a low-pressure head of 1 to 2 mm Hg is used to close the valve leaflets and therefore stress the leaflet material very little. In the strictest sense, however, stress-free fixation should mean that no external or internal forces are applied to the valve leaflet material. Because the valve cusps are composite structures containing a collagenous fibrosa, elastic ventricularis, and a loose, watery spongiosa.? none of these components would be stressed if zero-pressure fixation were truly stress-free. Altered collagen morphology and leaflet mechanics are
The Journal of Thoracic and Cardiovascular Surgery Volume 106, Number 2
the undesirable consequences of fixing valves under hydrostatic preload, Aortic valve cusps are constructed primarily from crimped collagenfibers/? that can stretch readily when the valve is pressurized with a hydrostatic head as low as several millimeters of mercury.'" ? This is thenatural mechanism that enables the leafletsto stretch, coapt, and seal off the ventricle during early diastole. In the preparation of bioprostheses, the valves are usually fixed closed under a hydrostatic pressure head of fixative. In such a scheme,the collagen fibers becomecrosslinked together with their crimp stretched out and do not recoil when the hydrostatic pressure is removed.Consequently, theleaflets losetheir complianceand respondvery differently from natural, unfixed valves, such as aortic valve allografts," New fixation techniques have therefore been designed to preserve natural collagen fiber crimp geometryand leaflet mechanics by reducing leaflet stress during fixation." Zero-pressure fixationtherefore has implicit advantages over low-pressure fixation. When zeropressure fixation is presented as stress-free fixation, however, a confusion ofterminology results. Ifzero-pressure fixation is truly free of stress, then zero-pressurefixed tissuewould have a collagencrimp and compliance very similar to those of fresh tissue. Althoughsome tensile testing studies have found little difference betweenfresh and zero-pressure-fixedporcine aortic valves," a deeper look into leaflet micromechanics isrequiredto fullyevaluate these newvalveconcepts.I D- 13 We know that the aortic valve contains both elastin and collagen-two proteins that react differently to glutaraldehyde fixation. The leaflets themselves are highly complex and wellorganized and contain three distinct fibrous layers: the fibrosa, the ventricularis, and the spongiosa/ (Fig. I). The fibrosa, a thicker, collagen-dominatedlayer, is believed to resist most of the tensile forces during valve closure. The ventricularis, on the other hand, contains both elastin and collagen- 15 and is highly extensible in the radial direction. In our previous work,13,16 we have shown that many of the wrinkles and folds in the fibrosa are maintained by virtue of its attachment to the ventricularis. When these two layers were separated, the fibrosa spontaneously straightened out and the ventricularis shrank. Some internal forces must have therefore been responsible for holding the fibrosa in its folded configuration. Internal stressessuch as these are not uncommon in biologic tissues. Fung and associates'" 18 have reported residualstressesin arteries and trachea and have tested for these stressesby selectively cutting the material and observing the retraction that takes place. If a similar phenomenon exists in the aortic valve leaflet, then fixing valves at zero pressure cannot guarantee truly stress-free fixation.
Vesely, Lozon, Talman
289
Fig. 1. Cutaway through leaflet and aortic wall showing internal configuration of fibrosa, spongiosa, and ventricularis. Nearthecentral portion ofleaflet, fibrosa exists asa folded sheet that unfolds as the leaflet stretches radially. We have therefore hypothesized that residual tensile and compressive stresses do exist within the fibrosa and the ventricularis of aortic valveleaflets. If this is the case, such stresses will change collagen fiber crimp geometry and the extensibilityof the leaflet layers after fixation. A clearer understanding of aortic valve structure and function at the fibrous level and a better definition of leaflet stressesduring fixationshould enable better evaluation of new valve designs. Methods
Thisstudy consisted of two parts. In the first part, wemeasured therelative amounts ofresidual strain that exist within the components of aortic valve cusps. Thiswas assessed byopticallymeasuring thedimensions of fibrosa andventricularis before andaftertheirseparation. Afterthe presence ofresidual strains was established, wemeasured theeffects oftheirpresence onthe extensibility ofthevalve leaflet components during fixation. This effect was assessed by means of uniaxial tensile tests. Microdissection. Whole hearts were obtained from freshly slaughtered pigs at the abattoir. These pigs ranged in agefrom 3 to 6 months and weighed an average of 100 kg. The hearts were transported to the laboratory, and the aortic valve cusps were cut out and kept in chilled pH-balanced saline solution during handling. The cusps were placed on a plastic dissecting
290
Vesely, Lozon, Talman
The Journal of Thoracic and Cardiovascular Surgery August 1993
Fig. 2. Microdissection process. By gently grasping ventricularis with dissecting forceps and pulling upwards, we exposed connecting fibers between the two layers, which could be severed with microscissors. In most cases, continuous pulling on tissue with instruments enabled layers to be torn apart along spongiosa, apparently the natural cleavage plane. board with the ventricularis side up. By gently grasping and lifting the ventricularis, we were able to expose the fibrous connections that spanned the spongiosa and connected the two layers. By carefully pulling on the ventricularis with blunt forceps, we separated the two layers along their natural cleavage plane, which appeared to be the midline of the spongiosa. Where such separation was not possible, for example, near the edges of the cusp, the fibers that held the layers together were cut one by one with microscissors. Such a process enabled the ventricularis to be separated from the fibrosa with minimal tearing (Fig. 2). Saline solution was applied repeatedly to keep the tissue moist. All dissection was performed with the use of a stereo microscope and, after dissection, the tissue samples were examined at the highest magnification available to ensure that the strips to be tested were not damaged in any way. This technique was slightly modified according to the size of the test specimens, as described later. Part 1. Optical measurement of dimensional changes. If the fibrosa and the ventricularis are under mutual compression and tension, as we have suggested.P: 16 then a change in dimensions should occur when these layers are' separated and the stresses are relieved. We therefore trimmed the whole leaflets into rectangular specimens with the cut edges oriented along the ridges visible on the aortic surface. The axes of the specimens were oriented in the standard radial and circumferential directions, and the specimens were then transferred to glass slides by floating them in saline solution and allowing them to settle onto the slides. In this way, the tissues were handled for a short time, and their natural dimensions were affected as little as possible. The slides were placed under a dissection microscope and a Sony
CCD video camera (model AVC-D?, Sony of Canada Ltd. Willowdale, Ontario) was connected to the auxiliary image port; an image of the rectangular piece of tissue was acquired by means of an image analysis package (JAVA, Jandel Scientific, Corte Madera, Calif.). These images were saved on disk on a personal computer for later analysis. The rectangular strip was then separated into its two layers as described previously, the fibrosa and ventricularis were floated in saline solution and deposited on slides by means of the method previouslydescribed, and similar images were acquired and saved. A calibrated scale was placed beside each specimen for the determination of the true dimensions regardless of the magnification of the microscope (Fig. 3). After images of all specimens were acquired, their dimensions in the radial and circumferential directions were measured with JAVA software. Because of the limited precision in cutting the specimens perfectly square, the accuracy of the dimension measurements is unlikely to be better than O.5mm. Part 2. Changes in extensibility. Ifthe fibrosa and the ventricularis are prestressed and the collagen fiber crimp is altered, then leaflet components fixed in glutaraldehyde under these conditions should have mechanical characteristics different from those of leaflet components fixed truly stress free. We assessed the effects of residual prestressing on the extensibility of porcine aortic valve leaflet components by examining the fibrosa and ventricularis individually. A total of three groups of tissue were studied. The first group consisted of fibrosa and ventricularis removed from aortic valve cusps that had been fixed at zero pressure. These tissues were therefore exposed to the naturally occurring internal leaflet stresses. The second group
The Journal of Thoracic and Cardiovascular Surgery Volume 106, Number 2
Vesely, Lozon, Talman
29 I
Fig. 3. Sections of valve cusps cut to measure elongation after separation of fibrosa and ventricularis. Whole tissue is shown on left and fibrosa alone on right. Note expansion of fibrosa in radial direction. Scale was imaged with tissue and used for calibration. consisted of fibrosa and ventricularis fixed in glutaraldehyde after they were separated from each other. These tissues therefore had no stresses imposed on them whatsoever. The third group included the same fresh specimens tested immediately after dissection. After testing, these fresh tissues were allowed to relax for 24 hours in a saline bath before fixation in glutaraldehyde. The fresh and separately fixed tissues acted as matched controls for each other. The three groups were termed ( 1) wholeor intact-fixed, (2) glutaraldehyde-treated, and (3) fresh fibrosa and ventricularis. Each of these was tested in the radial and circumferential directions. Ten specimens were in each group, for a total of 120 specimens. The fixative used in all cases was 0.5% glutaraldehyde buffered in phosphate buffer (0.067 moll L) at pH of 7. All tissues were fixed for a minimum of 24 hours before any handling or testing. Tensile testing. The three groups of tissues described previously were cut into 5 mm wide circumferential and 10 mm wide radial strips (Fig. 4). This was done to ensure that the width of the final strips was consistent, regardless of the degree of expansion or contraction that occurred during the dissection process.The extensibility of these strips was then measured with an Instron tensile testing machine (model 1125, Instron Corp., Canton, Mass.). The precision in cutting the strips to width was estimated to be ± 0.2 mm. Thickness was measured by a special device constructed for this purpose, II and the gauge length of the test strip was determined with a calibrated scale attached to the tissue-clamping mechanism. The precision of the thickness measurements and the gauge length were estimated to be ± 0.05 mm and ± 0.25 mm, respectively. The strips of tissue
.-_ •
• : ••
•• : ••
_-_._--------~
Circumferential
Radial
.... _---------_ .....- • •• Fig. 4. Whole leaflet or layers after dissection ~_
• .1
(top), before cutting specimens for tensile testing. Leaflet was trimmed to 5 mm wide circumferential strips and 10 mm wide radial strips, with circumferential and radial directions as shown. Nodulus Arantii was avoided in test strips.
292
The Journal of Thoracic and Cardiovascular Surgery August 1993
Vesely, Lozon, Talman
water level
..
I I....-
_ -
- - rod
_ - clamp
!-"...~ -
clamp
Side view Tissue straight
Tissue wrinkled
Tissue stretched
"0 C'O
o
...J
®
CD Buoyancy slope
Elongation
o.z rnm uncertainty
Fig. 5. Technique used to measure zero-load length of specimens. During first phase, only wrinkles were eliminated from tissue, as in 1 and 2. Load increased only because of decrease in buoyant force on clamp assembly as its parts were withdrawn from water. When tissue was completely straight and offered resistance, curve deviated from straight line as in 3. We then began to measure extensibility of specimens. Choosing gauge length at arbitrary level of load would have underestimated extensibility of specimens. were clamped between stainless steel jaws lined with No. 320 emery cloth to prevent slippage. The tissue was preconditioned by repeated stretching at an elongation rate of 10 mm/rnin to a maximum load of 0.98 N (100 gm) until reproducible load/ elongation curves were attained. The load/elongation curves were transformed to engineering stress versus strain curves by normalizing with the initial, unstrained dimensions of the specimens. All testing was done in physiologic saline solution heated to 37° C. Gauge length measurements. Because many of our conclusions for this study are based on the measured extensibility of the tissue strips, we took particular care in determining the starting length (gauge length) of the specimens at loads as close to zero as possible. The measurement of stress at some arbitrary value, for example, 0.5 gm, would produce consistent results but would prestretch the tissue strips, overestimate the gauge length,
and underestimate the extensibility. For the determination of the zero-load length, the tissue was clamped between the grips of the tensile testing machine, and the grips were moved closer together until the tissue was slightly wrinkled. The gain on the load amplifier was set at the maximum level, and the grips were slowly moved apart (Fig. 5). As the wrinkles in the tissue straightened out, no increase in tension was recorded at first. The only change was a slight increase in load as the tissue grip shaft was withdrawn from the water bath; this resulted from a slight decrease in the buoyant force on the grip shaft. During this phase, the unwrinkling of the tissue did not contribute enough force to affect the linear increase in load. After the tissue straightened and began to generate tensile resistance, the plotted curve sharply deviated from the straight line. This was the point at which we chose to measure the gauge length. The uncertainty in the measurement of gauge length was estimated
The Journal of Thoracic and Cardiovascular Surgery Volume 106, Number 2
Vesely, Lozon, Talman
29 3
o+--------,--~==:::::;======----.------------J
o
20 strain in %
10
30
40
Fig. 6. Method to calculate extensibility of test strip--the point at which a tangent drawn to the stress/strain curve at 300 kPa crossed the x axis, as shown. Extensibility was expressed as % strain. In this example, the extensibility is 31%.
Table I. Mean changes in length of leaflet components after their separation Change in dimensions (%) Tissue Fibrosa Ventricularis
Direction
No.
Mean
SD
Circumferential Radial Circumferential Radial
II II II II
- 4.78 30.47 -13.37 -12.29
5.11 13.19 2.74 3.75
A negative number denotes contraction, and a positive number denotes expansion. All values are change in percentage relative to dimensions when the two layers were connected. SD, Standard deviation.
to be ± 0.1 mm. For the shortest specimen, with a gauge length of only 7 mm, our estimate of gauge length would have an error of only 1.4%. In our view, this technique is the least likely to underestimate the extensibility of highly extensible tissue, especially when compared with other techniques that measure gauge length at an arbitrary load. This technique is therefore particularly good for analyzing the initial, high-extensibility region of the stress/strain curve of radial strips of aortic valve cusp tissue. We admit, however, that our technique can overestimate tissue extensibility if the tissue is not positioned squarely between the tissue grips. After much practice, however, this effect was kept to a minimum. This protocol has been used in our laboratory for several years and has produced consistent results. The collagen fiber crimp was assumed to be proportional to the extensibility of the tissue strips, which was measured from
Fig. 7. Inward contraction of specimens immediately after dissection caused by tension in ventricularis released by cut. Contraction produced slight bowing of specimen and skewing of cut face inward.
the length of the low-modulus phase of each stress/strain curve. This was obtained from the intersection of the x axis with a line drawn tangent to the curve at a stress of 300 kPa (Fig. 6). The extensibility was expressed as percentage of strain. A change in
The Journal of Thoracic and
294
Vesely, Lozon, Talman
Cardiovascular Surgery August 1993
Mean and SEM Circumferential Fibrosa
Radial Fibrosa Circumferential Ventricularis
Radial Ventricularis
o -20
-10
o
10
20
30
% change in length
•
Circumferential Fibrosa
•
10
20
30
Radial Fibrosa
~ Circumferential Ventricularis ~ Radial Ventricularis
Fig. 8. Mean changes in length of fibrosal and ventricularis leafletcomponents after separation.Error bars denotestandard error of mean. Slight contraction of fibrosa in circumferential directionwas unexpected and likelyresulted from realignment of fibers rather than major contractionof collagen fibercrimp. extensibility would indicate to what extent the collagen crimp wasaffectedby the fixation process. Becauseall specimens were analyzed with the same method,our resultsare internallyconsistent, and comparisons between groups are valid. Results General observations. The prestressing of the fibrosa and the ventricularis became apparent as soon as rectangular specimens were cut from the leaflets and observed at high magnification. The cut edges of the leaflets were not square but were skewed toward the ventricularis, as shown in Fig. 7. This occurred when a segment of tissue was cut and separated from surrounding tissue. Because no loads could exist at the edges of the tissue, the edges moved laterally in response to the stress field in the interior of the specimen. The ventricularis therefore contracted slightly, and the fibrosa expanded. For consistency, the dimensions of the fibrosa side were used as the dimensions for the whole tissue. Part 1. Dimension changes. The mean dimensions of the rectangular specimens before dissection were 8.37 ± 1.91 mm in the circumferential direction and 6.10 ± 1.53 mm (mean ± standard deviation) in the radial direction. After separation of the two layers, the fibrosa expanded considerably (by 30%) in the radial direction. This was statistically significant (p < 0.00001). In the circumferential direction, however, a small but statistically significant shrinkage (less than 5%) was observed after dissection (p < 0.02; Fig. 8, Table I). The ventricularis, on the other hand, contracted radially by
40
50
60
70
Strain in %
40
o
Fresh
•
Treated
•
Whole
Fig. 9. Mean extensibility of fibrosa and ventricularis, tested in radial and circumferential directions in freshand glutaraldehyde-treatedstates. Treated denoteslayersthat havebeenfixed whileseparated, and whole denotes that they have been fixed whilestill connected together. Statistical comparisons between multiplegroups are shownin Table III. Note increased extensibility of fibrosa from whole-fixed leaflets in both radial and circumferential directions, compared with both fresh and separately-treated specimens. Error bars denotestandard error of mean. 12% (p < 0.00001) and circumferentially by 13% (p < 0.00001). Because the leaflet components spontaneously changed their length after separation from each other, the fibrosa and the ventricularis must be naturally preloaded; most notably, the fibrosa was preloaded in the radial direction. This preload can be considered residual stress that may affect tissue behavior after fixation. Part 2. Tensile testing. The effects of prestressed and stress-free fixation on the extensibility of the leaflet components were evaluated from the changes in the measured extensibility of rectangular test strips. The radial extensibility of the fibrosa did not change when it was fixed separately, but increased significantly when it was fixed as part of the whole leaflet (39% whole-fixed versus 29% fresh;p < 0.0122) (Fig. 9, Tables II, III). A similar effect occurred in the circumferential direction. The separated fibrosa, fresh and treated leaflets, had similar extensibilities (8% and 7%, respectively,) but fibrosas from leaflets fixed whole were more extensible (13%, P < 0.0003 or better). An opposite effect occurred in the radial ventricularis. Ventricularis from the whole-fixed leaflets had a significantly decreased extensibility when compared with fresh leaflets (35% whole-fixed versus 64% fresh; p < 0.00001). There was no statistically significant difference in the extensibility of any ventricularis specimens in the circumferential direction.
The Journal of Thoracic and Cardiovascular Surgery Volume 106, Number 2
Vesely, Lozon, Talman
295
Table II. Extensibilities offibrosa and ventricularis Extensibility (%) State Fresh
Layer Fibrosa Ventricularis
Treated
Fibrosa Ventricularis
Whole-fixed
Fibrosa Ventricularis
Direction
No.
Mean
SD
Circumferential Radial Circumferential Radial Circumferential Radial Circumferential Radial Circumferential Radial Circumferential Radial
10 10 10 10 10 10 10 10 II 9 10 10
8.150 29.500 13.460 63.660 7.140 33.150 19.420 57.810 12.750 39.230 14.650 35.410
2.533 9.417 5.450 11.260 2.916 8.243 6.964 9.096 2.337 5.036 5.025 6.461
Treated, Layers that were fixed while separated; Whole fixed, layers that were fixed while connected. See also Fig. 9. SD, Standard deviation.
In summary, when leaflet layers were separated before glutaraldehyde fixation, no significant changes in extensibility were observed, but leaflet layers fixed together did show changes. This occurred both radially and circumferentially in the fibrosa and radially in the ventricularis. Apparently, the presence of residual stresses within aortic valve leaflets during zero-pressure fixation significantly changed the extensibilities of both the fibrosa and the ventricularis components. Discussion It has been shown that reducing fixation pressures lowers leaflet stresses, minimizes the distortion of the collagen fiber crimp.> 7 and preserves a significant portion of the natural mechanical function of aortic valve cusps. The term zero-pressure fixation, however, has sometimes been linked with the term stress-free fixation. Such a linkage suggests that zero-pressure-fixed valves have unchanged collagen fiber crimp structure and therefore distensibility and compliance characteristics similar to those of fresh tissue. This has been demonstrated to some extent with standard biaxial tensile testing." Recent work in basic biomechanics, however, suggests that most biologic structures, including the aortic valve, are prestressed even in the relaxed state. 13, 16, 18 In this study, we have demonstrated the existence of residual stresses in the fibrosa and the ventricularis and have shown that even zero-pressure fixation does not guarantee truly stress-free fixation. Certain assumptions were made in this study. In part 1, we assumed that the dimensions of the leaflet layers were not altered significantly during handling and that the differences in size observed between the layers after
Table III. Statistical comparisons among the groups shown in Table II Groupings
Comparisons
Extensibility
Circumferential p < 0.00001 vs radial Ventricularis Circumferential p
Fibrosa
NS, Not significant. Level of significance was determined to be 0.05 by means of a small sample t test. Paired t tests were used to compare fresh and treated groups because they were the same tissue.
29 6
Vesely, Lozon, Talman
dissection were not artifactual. We believe that if great care is taken in floating the leaflets onto glass slides, the only force present that could distort the dimensions of the leaflets would be surface tension, and we believe this distortion to be minimal. In part 2, we assumed that the changes in extensibility measured with uniaxial tensile tests were real and that similar changes would exist in the whole aortic valve leaflets when stretched biaxially during normal cardiac function. We should point out, however, that uniaxial tensile tests tend to overestimate tissue extensibility. This results from fiber reorientation as the strip is stretched in one direction only. When stretched biaxially, these fibers reorient much less and offer resistance to stretching at lower extensions. The extensibility values presented in this article were therefore used only as a comparison between groups and do not represent the true extensibility that would be expected from a whole leaflet. We chose uniaxial over biaxial testing because the technique for determining gauge length and measuring the very high extensibility region of the stress/strain curve cannot be readily applied to biaxial testing machines. We also assumed that the measured extensibility of tissue is determined largely by the state of the collagen crimp. In reality, extensibility is only an estimate of collagen crimp because some of the extensibility occurs by means of a realignment of collagen fibers and not as a result of uncrimping. Finally, we assumed that glutaraldehyde fixation crosslinked the collagen fibers in the configuration that existed during the treatment process. If the fibers were uncrimped during fixation, they would have stayed uncrimped, even after the load applied to them during fixation was removed. This enabled us to measure differences inextensibility before and after fixation and to relate these differences to the loads that were present during fixation. Although this assumption is valid for the fibrosa, which does not appear to have much elastin, it may not be entirely valid for the ventricularis. Elastin is immune to cross-linking with glutaraldehyde'? and would therefore attempt to recoil to its original configuration after the load had been removed. Ventricularis fixed in the whole leaflet might recoil and contract after being separated from the fibrosa. Although we did not specifically test for this, we did not observe any recoil during dissection of the wholefixed cusps, as we did with the fresh cusps. It is therefore possible that the extensibility of the ventricularis from leaflets fixed whole may be overestimated because of this phenomenon. Nevertheless, because we have shown that whole-fixed ventricularis has less extensibility than does fresh or fixed ventricularis, the possible recoil of the whole-fixed ventricularis does not alter our conclusions.
The Journal of Tboracic and Cardiovascular Surgery August 1993
In the context of the previous assumptions, our results provide additional insight into the internal micromechanics of aortic valve leaflets and the effects of conventional glutaraldehyde-fixation techniques. The failure of existing bioprosthetic valves to provide satisfactory service and durability may be related to our limited knowledge of leaflet micromechanics and the glutaraldehyde-fixation process itself. Some of the pioneering work by Broom and associates'v? has demonstrated that valve tissues fixed under high pressures do not have natural mechanical function and may therefore wear out prematurely. Our earlier studies 10, II have suggested that the failure of glutaraldehyde-fixed tissue may occur as a result of flexural damage to the fibrous components of the leaflet material. Unfixed materials, when bent, do not exhibit the sharp kinking and compressive buckling of fixed tissues. 10, 20 If mechanical damage brought about by unnatural mechanical characteristics does indeed contribute to bioprosthetic valve failure, then such mechanical damage can be minimized by means of properly controlled glutaraldehyde-fixation techniques. Understanding the function of the valve leaflets at the fibrous level is therefore a prerequisite to developing a fixation process that can preserve the natural structure and function of the valve leaflet material. We now know that even internal leaflet stresses during fixation affect valve function at the micromechanicallevel. Because the fibrosa taken from a whole-fixed cusp is more extensible than one that is fixed alone, the fibrosa must have been under radial compression by virtue of its attachment to the ventricularis. The "freezing" of these fibers during fixation under compression increased the amount of radial fiber crimp and therefore increased their radial extensibility. In the ventricularis, this phenomenon occurred in the opposite direction. Because the ventricularis from whole-fixed cusps had a lower extensibility than either the fresh or fixed tissue, the ventricularis must have been under tension during fixation. Tension reduces the collagen fiber crimp and the resultant extensibility. This residual tension in ventricularis may also explain why the leaflet has a tendency to flip and curl toward the ventricularis after it is cut free from the aortic root. In most cases, the changes in extensibility measured with uniaxial tensile testing reflected the preload, which could be inferred from the changes in dimensions after dissection. The fibrosa was under compression in the radial direction and was naturally expanded when released from the ventricularis; it had an increased collagen fiber crimp when fixed while part of the whole leaflet and therefore had an increased extensibility. The ventricularis, on the other hand, was under tension in the radial
The Journal of Thoracic and Cardiovascular Surgery Volume 106, Number 2
direction, contracted when released from the fibrosa, had a decreasedcollagen fiber crimp when fixedas part of the whole leaflet,and therefore had a decreased extensibility. A similar scenario occurred for the ventricularis in the circumferential direction but not for the fibrosa. The circumferential fibrosa was expected to expand circumferentiallywhen dissected free but, instead, shrank slightly. Although the shrinkage was less than 5%, it was statistically significant. This finding remains unexplained but may be related to the interactions between collagen fiber crimp and fiber migration during tensile testing, as mentioned previously. These interactions are still not well understood. What does zero-pressure fixation really mean? In this study, we have concluded that zero-pressure fixation does not ensure truly stress-free-fixed valves. Although there may not be any external forces applied to the valve cuspsduring fixation, internal forces are present by virtue of the interactions between the fibrosa and the ventricularis. When the whole valve is fixed, even under zero transmural pressure, the ventricularis is fixed under tension and the fibrosa is fixed under compression. Because the crimp of the collagen fibers is "frozen" into the tissue during glutaraldehyde fixation, zero-pressure-fixed valves cannot have the same mechanical behavior as the fresh valves. Although the wide-ranging effects of this typeof fixationare still not wellunderstood, zero-pressure fixation stillaffects leaflet mechanics much less than lowpressure fixation.P Further confusion may arise from reference to leaflet pressure and transmural pressure. Recently,the importance of aortic root geometry has been recognized." and some valvesare being fixed at high root pressures to simulate diastolic root geometry while still maintaining pressures across the leaflets that are very close to zero. Such valves can be correctly called zeropressure-fixedvalvesif the differencesbetween the stresses within the leaflets and the aortic wall are acknowledged. New fixation schemes. Because it has been shownthat stresses during fixation reduce collagen fiber crimp and produce leaflets with unnatural mechanical function.v? the control of fixation pressures has been an important focusof many heart valve manufacturers. More recently, however, it has been suggested that other leaflet micromechanics,in addition to collagen fiber crimp, are important in defining physiologic valve function. 10, II, 14,22 The interaction of the fibrosa and the ventricularis is one such critical issue.P: 16 More innovative fixation schemes may have to be applied to preserve the natural micromechanical function of aortic valve cusps. Dynamic fixation, a technique first proposed by US,22 has been shown to
Vesely, Lozon, Talman
297
improve some of the mechanical characteristics of aortic valve tissue.v' Perhaps the best type of fixation is none at all. In the future, it may be more appropriate to attempt to eliminate the problem of antigenicity of xenograft material by means of techniques other than glutaraldehyde fixation and to use a more direct approach, such as antigen extraction and masking.i" We are grateful for the assistance of Dr. D. R. Boughner in the composition of this manuscript. REFERENCES I. JaffeWM, Barratt-Boyes BG,SadriA, Gavin JB, Clover-
dale HA, NeutzeJM. Earlyfollow-up of patients with the Medtronic Intact porcine valve. J THoRAc CARDIOVASC SURG 1989;98: 181-92. 2. Gross L, Kugel MA. Topographic anatomy and histology ofthe valves in the humanheart.Am J Pathol1931 ;7:44556. 3. Clark RE, Finke EH. Scanning and light microscopy of human aortic leaflets in stressed and relaxed states. J THORAC CARDIOVASC SURG 1974;67:792-804. 4. HilbertSL, Ferrans VJ, Swanson WM. Optical methods forthenondestructive evaluation ofcollagen morphology in bioprosthetic heart valves. J Biomed Mater Res 1986; 20:1411-21. 5. Hilbert SL, Barrick MK, Ferrans VJ. Porcine aorticvalve bioprostheses: a morphologic comparison of the effects of fixation pressure. J Biomed Mater Res 1990;24:773-87. 6. Broom ND, Marra D. Effects of glutaraldehyde fixation andvalve constraint conditions on porcine aorticvalve leaflet coaption. Thorax 1982;37:620-6. 7. Broom ND, Thomson FJ. Influence of fixation conditions on the performance ofglutaraldehyde-treated porcine aortic valves: towards a more scientific basis. Thorax 1979; 34:166-76. 8. Vesely I, Gonzalez-Lavin L, Graf D, Boughner DR. Mechanical testing ofcryopreserved aorticallografts: comparison with xenografts and fresh tissue. J THORAC CARDIOVASC SURG 1990;99:119-23. 9. Mayne ASD, Christie GW, Smaill BH, Hunter PJ, Barratt-Boyes BG. Anassessment ofthe mechanical properties ofleaflets from foursecond-generation porcine bioprostheseswith biaxial testing techniques. J THORAC CARDIOVASC SURG 1989;98: 170-80. 10. Vesely I, Boughner DR, Song T. Tissue buckling as a mechanism ofbioprosthetic valve failure. AnnThoracSurg 1988;46:302-8. II. Vesely I, Boughner DR. Analysis of the bending behaviour of porcine xenograft leaflets and of natural aortic valve material: bending stiffness, neutralaxisandshearmeasurements. J Biomech 1989;22:655-71. 12. Vesely I. Analysis of the Medtronic Intact bioprosthetic valve: effects of "zero-pressure" fixation. J THORAC CARDIOVASC SURG 1991;101:90-9.
298
Vesely, Lozon, Talman
13. Vesely I, Noseworthy R. Micromechanics of the fibrosa and ventricularis of aortic valve leaflets. J Biomech 1992; 25:101-13. 14. Vesely I, Krucinski S, Campbell G. Micromechanics and mathematical modelling: an inside look at bioprosthetic valve function. J Cardiac Surg 1992;7:85-95. 15. Jonas RA, Ziemer G, Britton L, Armiger LC. Cryopreserved and fresh antibiotic-sterilized valved aortic homograft conduits in a long-term sheep model. J THoRAc CARDIOVASC SURG 1988;96:746-55. 16. Vesely I, Lozon A. Natural preload of aortic valve leaflet components during glutaraldehyde fixation: effects on tissue mechanics. J Biomech 1993;26:121-31. 17. Fung YC, Liu SQ. Change of residual strains due to hypertrophy caused by aortic constriction. Circ Res 1989; 65:1340-9. 18. Han HC, Fung YC, Residual strains in porcine and canine trachea. J Biomech 1991;24:307-15.
The Journal of Thoracic and Cardiovascular Surgery August 1993
19. Fung YC, Sobin SS. The retained elasticity of elastin under fixation agents. J Biomech Eng 1981;103:121-2. 20. Broom ND. An in vitro study of mechanical fatigue in glutaraldehyde-treated porcine aortic valve tissue. Biomaterials 1980;1:3-8. 21. Vesely I, Menkis AH, Rutt B, Campbell G. Aortic valve/ root interactions in porcine hearts: implications for bioprosthetic valve sizing. J Cardiac Surg 1991;6:482-9. 22. Song T, Vesely I, Boughner DR. Effects of dynamic fixation on the shear behavior of porcine xenograft valves. Biomaterials 1990;11:191-6. 23. Mavrilas D, Missirlis Y. An approach to the optimization of preparation of bioprosthetic heart valves. J Biomech 1991; 24:331-9. 24. Vesely I, Noseworthy R, Wilson G. Development of a hybrid xenograft/autograft aortic valve bioprosthesis. Can J CardioI1991;7[suppl]83A.
Bound volumes available to subscribers Boundvolumesof THEJOURNAL Of THORACIC AND CARDIOVASCULAR SURGERY are availableto subscribers(only) for the 1993issues from the Publisher, at a cost of $75.00 for domestic,$98.25 for Canadian, and $93.00for international subscribersfor Vol. 105(JanuaryJune) and Vol. 106 (July-December). Shipping charges are included. Each bound volume contains a subject and author index and all advertising is removed. Copiesare shipped within 60 days after publicationof the last issueof the volume.The bindingis durable buckram with the JOURNAL name, volume number, and year stamped in gold on the spine. Paymentmust accompany all orders. Contact Mosby, SubscriptionServices,11830WestiineIndustrial Drive,St. Louis,Missouri63146-3318,USA; phone I (800) 453-4351 or (314) 453-4351. Subscriptions must be in force to qualify. Boundvolumes are not availablein place of a regular JOURNAL subscription.