ABSTRACT OPTIMIZATION OF ENGINEERED HEART VALVE TISSUE EXTRACELLULAR MATRIX George Carl Engelmayr, Jr., PhD University of Pittsburgh, 2005 Prosthetic heart valves, whether biologically-derived or mechanical, have improved the quality of life of millions of patients worldwide since their introduction in the 1960's. However, while currently available prosthetic valves perform sufficiently well in the short term, the side-effects of anticoagulation therapy (mechanical valves) and the structural degeneration of bioprosthetic and allograft valves represent significant drawbacks in the long-term. The limitations of these non-viable devices are especially pronounced in pediatric patients suffering from congenital valvular lesions, as none of the current prosthetic valves have the capacity to grow in tandem with the somatic growth of the patient. Tissue engineered heart valves (TEHV) are conceptually appealing for use in the surgical repair of valvular lesions because they harbor a living cell population potentially capable of orchestrating tissue self-repair, growth, and resistance to infection. Since the mid-1990's, significant progress has been made toward the development of a functional TEHV, culminating in long-term implantation studies in sheep. In these previous studies, TEHV were constructed by seeding vascular-derived smooth muscle and endothelial cells, or bone marrow-derived mesenchymal stem cells (BMSC), onto bioresorbable polymer scaffolds. The resultant TEHV were subsequently cultivated in a pulsatile flow loop bioreactor in which TEHV could be exposed to graduated increases in mechanical stimulation prior to implantation. Importantly, it was demonstrated through these previous studies that mechanical stimulation is critical to the development of a functional TEHV. In the absence of mechanical stimulation, TEHV exhibited significantly reduced extracellular matrix (ECM) formation, and thus upon scaffold degradation retained insufficient structural integrity for acute hemodynamic function. However, because the various mechanical stimuli (e.g., cyclic flexure, tension, and fluid flow) were coupled in the pulsatile flow loop bioreactor, it could not be deduced how the individual modes of mechanical stimulation contributed to the overall TEHV developmental response. Such information is essential, both for developing rationally designed mechanical conditioning regimens, and importantly for potential clinical applications, for quantifying the sensitivity of the tissue formation process to perturbations in these factors. To establish biomechanical end-points for evaluating TEHV, in the current study mathematical models were developed to predict the effective stiffness of TEHV biomaterials from the properties and structure of the individual constituents. It was found that the effective stiffness of the nonwoven polymer scaffolds could accurately be predicted from the spring-like tensile properties and orientations of the scaffold fibers, and that the primary mechanical effect of ECM deposition was an increase in the number of fiber-fiber bond points. Toward quantifying the independent and coupled effects of mechanical stimulation on TEHV development, a novel flex-stretch-flow (FSF) bioreactor was developed in which multiple TEHV specimens could be subjected to well-defined combinations of mechanical stimuli. Finally, the FSF bioreactor was used to elucidate the independent and coupled effects of cyclic flexure and laminar flow on TEHV tissue formation by BMSC. The combination of flexure and flow was found to synergistically accelerate tissue formation and BMSC differentiation, thus paving the way toward rational designs for TEHV conditioning regimens utilizing novel cell sources.

Following glenohumeral joint dislocation, surgical repair is often advocated where the glenohumeral capsule is shifted and plicated. However, nearly 25% of patients still experience redislocations. To improve these results, functional evaluations (experimental and computational) of the glenohumeral capsule have been performed whereby isolated, discrete capsuloligamentous regions of the capsule were examined. Specifically, the capsuloligamentous region termed the anterior band of the inferior glenohumeral ligament is often examined in this way since it is frequently injured during dislocations. However, this practice may not be appropriate as recent data suggests that the glenohumeral capsule functions multiaxially. Therefore, the objective of this work was to compare the predicted strain distribution and deformed shape of the anterior band of the inferior glenohumeral ligament to that experimentally measured for two finite element models: 1) composite model including all capsuloligamentous regions and 2) discrete model including only the anterior band of the inferior glenohumeral ligament. The average maximum principal strain for the anterior band of the inferior glenohumeral ligament was 21±14%, 35±14%, and 0±1% for the experimental measurements, composite finite element model, and discrete finite element model, respectively. Thus, the predicted strain distribution in the anterior band of the inferior glenohumeral ligament was similar to that which was experimentally measured for the composite finite element model. Additionally, the predicted deformed shape in the composite finite element model was also similar to experimental data with the anterior band of the inferior glenohumeral ligament clearly wrapping around the humeral head. However, the predicted strain distribution and shape for the discrete finite element model was drastically different from that observed experimentally with the anterior band of the inferior glenohumeral ligament twisting somewhat along its longitudinal axis and buckling away from the humeral head. These differences may be attributed to neglecting the boundary conditions along the margins of the anterior band of the inferior glenohumeral ligament applied by the remaining capsuloligamentous regions. Thus, the glenohumeral capsule should be evaluated as a sheet of fibrous tissue and composite finite element models may be utilized to evaluate its function in the normal, injured, and surgically repaired state.

Shape optimization has been used for decades to improve the aerodynamic performance of automobiles and aircraft. The application of this technology to blood-wetted medical devices have been limited, in part, due to the ambiguity of hemodynamic variables associated with biocompatibility specifically hemolysis, platelet activation, and thrombus formation. This study undertook a systematic evaluation of several objective functions derived directly from the flow field. We specifically focused on the optimization of a two-dimensional blood conduit (cannula) by allowing free variation of the centerline and cross-sectional area. The flow was simulated using computational fluid dynamics (CFD) at a nominal flow rate of 6 lpm and boundary conditions consistent with an abdominally positioned left-ventricular-assist device (LVAD). The objectives were evaluated both locally and globally. The results demonstrated similarities between four of the functions: vorticity, viscous dissipation, principal shear stress, and power-law (PL) blood damage models based on shear history. Of the functions analyzed, those found to be most indicative of flow separation and clearance were flow deviation index and the Peclet Number. The conclusions from these studies will be applied to ongoing development of algorithms for optimizing the flow path of rotary blood pumps, cannula, and other blood contacting devices.

Skeletal muscle injury is a common type of injury in sports medicine. After the injury, the traumatized muscle undergoes sequential and overlapped phases of healing, including degeneration, inflammation, regeneration, and fibrosis. Inflammation is an important phase in the natural healing process of many injured tissues. During this phase, various cytokines and cells participate and form a complex environment. Since uncomfortable symptoms are associated with inflammation, current treatments for skeletal muscle injury focus on inhibiting the inflammation phase by using non-steroidal anti-inflammation drugs (NSAIDs). Given that inflammation is shown to be beneficial to the healing process of many other tissues, it is necessary to study its role in skeletal muscle healing in order to improve the healthcare of sports-related injuries. Here, we investigated the role of the inflammation phase in skeletal muscle injury. We initially compared the healing of injured skeletal muscle and the growth of muscle cells with and without the treatment of NS-398, a cyclooxygenase-2 (COX-2) specific NSAID. We observed that NS-398 impaired muscle healing by delaying muscle regeneration and increasing scar tissue formation. NS-398 inhibited the proliferation and differentiation of muscle cells, and the expression of prostaglandin E2 (PGE2) and prostaglandin F2¦Á (PGF2¦Á). Next, we investigated the important components in the inflammation phase, including the COX-Prostaglandins pathway, transforming growth factor-¦Â1 (TGF-¦Â1), and macrophages. We found that PGF2¦Á and PGE2 may promote muscle cell maturation. The expression of TGF-¦Â1, a fibrotic growth factor, may be suppressed by PGE2 but increased by NS-398. Macrophages may promote muscle healing by increasing the release of various cytokines and growth factors, including TGF-¦Â1 and PGE2. These results suggest that various cytokine, growth factor, and cellular components participate in the inflammation phase and cooperate with each other to modulate the healing process. Simply using NSAIDs to block inflammation may not be the optimal treatment in the effort of achieving complete recovery after muscle injury. In conclusion, our results suggest that the inflammation phase is important in skeletal muscle healing and further research to investigate the role of its components is necessary.

The purpose of this study was to improve the efficacy of VAD therapy for patients intended for VAD insertion. The study focused on the specific decision whether an LVAD or BiVAD is appropriate. A hierarchical decision model was constructed using an influence diagram of clinical risk factors derived through interviews with expert cardiologists and cardiac surgeons. Most of the variables are summarized by two independent criteria: risk of surgery and risk of right ventricular (RV) failure. These risks are computed from various patient demographics, tests, and hemodynamics using expert physician-selected weighted linear and weighted non-linear relationships. The model was validated with retrospective data from patient records at University of Pittsburgh Medical Center (UPMC) for patients implanted after 1990 and explanted before 2006. In total 239 patients were implanted and explanted during this time, of those 168 had sufficient information to be used in this analysis. 48 patients received biventricular assistance (BiVADs), 119 patients received only left ventricular assistance (LVADs). Of these 119 LVAD patients, 19 subsequently received an RVAD due to unanticipated RV dysfunction. Pre-implant data were used as input to the model. The model parameters were derived from two different physicians. The models based on individual physicians weightings predicted 63% (47%) of the patients who required an RVAD after implant. However, these decision models also recommended BiVAD implantation for 40% (43%) of patients who were treated successfully with an LVAD alone. A nonlinear numerical optimizer was used to improve the model parameters to optimize the agreement with eventual outcomes. The optimized model predicted 74% of the patients who required an RVAD post-implant and recommended the implantation of BiVADs in 21% of patients who were treated successfully with an LVAD alone. In conclusion, the decision model provided a more aggressive use of biventricular assistance, which retrospectively would have benefited patients who required an RVAD at a later date, but would have unnecessarily implanted RVADs in some patients that survived with an LVAD alone. However the model also identified that 48% of the patients who initially received BiVADs to be candidates for LVAD alone.

It has been established that information from the visual, somatosensory, and vestibular systems contributes to balance, but how this information is integrated remains unclear. Sensory integration is a temporally dynamic process in which the sources for sensory information are dynamically regulated and change as environmental conditions change. Significant differences in this dynamic regulation have been found among healthy young and old subjects, as well as subjects with vestibular disease. The present research was designed to examine the impact of aging and unilateral vestibular disease on balance as subjects responded to rapid changes in visual and somatosensory input. The postural sway of 25 healthy young controls, 24 healthy older controls, and 7 older subjects with unilateral vestibular hypofunction (UVH) was measured while visual and proprioceptive transitions were induced. To produce a sudden change in the visual environment, the amplitude of a sinusoidally moving visual scene was rapidly increased. Somatosensory information was altered through the use of a support platform that rotated in proportion to body sway, thereby reducing information from the somatosensory system. The power of anterior-posterior head velocity was calculated for the 20 s surrounding each transition. This segment was then broken into 5 s periods in order to investigate adaptation, i.e. within-trial time-varying characteristics of postural sway. Habituation was studied by investigating the changes in postural sway over repeated trials. A mixed factor repeated measures ANOVA was conducted with the power of postural sway velocity (dB) as the dependent variable. The independent variables were trial repetition and the time period in relation to the stimulus transition. Subject type was the between-subjects factor. Both healthy and vestibularly impaired older subjects were observed to sway more than healthy younger subjects during all experimental conditions. Following a decrease in reliable somatosensory input, all subjects showed an increase in postural sway power. This increase was greatest in older subjects with UVH. Though adaptation following the perturbation was seen in all subject groups, this process was slower in the patient group. Habituation was seen in most trial conditions, especially between the first and second presentations of a stimulus.

Bleeding and thromboembolism remain major complications of ventricular assist device (VAD) support. The amount of biocompatibility information that may be collected during preclinical studies is limited due to a lack of available assays, leaving the evaluation of investigational devices incomplete. To address these issues, flow cytometric assays were developed to quantify bovine circulating activated platelets, platelet microaggregates, platelet-leukocyte aggregates, and monocytes expressing tissue factor. Platelet lifespan was determined using an ex vivo biotinylation technique. These assays were applied in 50 animals receiving the Nimbus/Heartmate II axial flow VAD, 29 receiving the SunMedical EVAHEART centrifugal VAD, over 20 animals receiving a variety of other cardiovascular devices, and eight animals that underwent a sham VAD implantation procedure. The results demonstrated significantly increased circulating activated platelets and leukocytes, and cell aggregates following VAD implantation, which then usually declined to a lower but still significantly elevated level. Deviations from this pattern were observed in several pumps with obstructive thrombi in the blood flow path. Platelet life span decreased and platelet consumption correspondingly increased. The sham studies demonstrated that the effects of the implant procedure abated within three weeks. Thus, the ongoing platelet and leukocyte activation and aggregation, and decreased platelet life span could be attributed to the VADs, even while accounting for surgical effects. To identify the potential causes of the observed cellular activation, VAD surface modifications, revolutions per min increases, and anticoagulant regimen changes were evaluated in vivo. Two blood-shearing devices were constructed to investigate the effects of the supraphysiologic shear field within rotary VADs, although heat generation and sealing issues limited their effectiveness. Flow visualization of the Heartmate II VAD revealed vortices developed at low flow rates, frequently encountered in vivo. In conclusion, the propensity of cardiovascular devices to activate platelets and leukocytes was quantified, while accounting for the effects of the implant procedure. Through in vivo and in vitro investigations, it was demonstrated that the blood-contacting surface and adverse flow effects each contributed to the observed cellular activation. Thus, applying novel biocompatibility assays to preclinical studies, including those evaluating design enhancements and refinements, may be used to develop safer cardiovascular devices.

Stable upright stance is achieved through an active postural control process that requires the accurate integration of sensory feedback signals from the visual, graviceptive and proprioceptive systems. Previous studies have shown that this integration process may involve reweighting, whereby the relative contributions of the various sensory signals are dynamically altered in order to minimize reliance on unreliable signals. In addition, evidence suggests that feedback control by itself cannot explain experimental observations of postural behavior. In light of these observations, the current study proposed that a predictive mechanism exists within the postural control system that can identify the highly repetitive pattern within a predictable sensory input, and can use this information to facilitate the sensory reweighting process. The objectives of the current work were to: (1) uncover experimental evidence of such a predictive mechanism, through examination of postural sway responses in healthy young adults exposed to various types of predictable and unpredictable moving visual scenes; (2) examine how this predictive mechanism manifested itself in individuals who were particularly reliant on visual sensory information; and (3) determine if this predictive mechanism was influenced by cognitive tasks, which are thought to interact with the sensory reweighting process. Data revealed that in healthy young adults predictable stimuli elicited improved sway performance compared to unpredictable stimuli, as indicated by significant decreases in both overall sway magnitude, and the time required for sensory reweighting to occur. This effect was enhanced during the performance of a concurrent cognitive task, but was not observed in visually dependent individuals, apparently due to an inability to perform sensory reweighting. Taken together, these observations support the existence of a predictive component to postural control that can alter the dynamic reweighting of sensory inputs during exposure to predictable stimuli. These findings may have implications for the design of experiments involving moving visual scenes, as well as for the treatment of individuals suffering from certain balance disorders.

Approximately 250,000 400,000 individuals in the United States have spinal cord injuries (SCI); urologic complications, including bladder dysfunction, are among the most common problems these patients encounter. Although extensive studies have been conducted on the effects of spinal cord injury on bladder function, the alterations in mechanical behavior and functional properties of the bladder wall tissue and the underlying mechanisms are not well understood. Using a rat model of SCI, it has been previously demonstrated that the bladder wall significantly remodeled in early stages after injury. The remodeling process included changes in mechanical properties, composition and structure of the bladder wall, and occurred as early as 10 days post-injury. Based on the previous findings, it was hypothesized that the altered mechanical environment of the urinary bladder following spinal cord injury was the key signal for the changes in the tissue functional properties. In order to test this hypothesis and gain a better understanding of relationship between function and structure of bladder wall following SCI, the present study combined different experimental methods (including mechanical testing, biochemical assays and histomorphometry) to investigate changes in mechanical properties, as well as alterations in composition and morphology of the bladder wall tissue at various time points up to 10 weeks after injury. Changes in mechanical compliance and material class found during the biomechanical analyses clearly indicated that the bladder wall continuously remodels after spinal cord injury beyond the time point previously tested. The results of histomorphometric study provided first evidence that bladder smooth muscle cells exhibited hypertrophy rather than hyperplasia, corroborated the previous mechanical anisotropy data, and provided the basis for development of structure-based constitutive models for urinary bladder wall tissue. Finally, the findings of biochemical study demonstrated that changes in extracellular matrix of bladder tissue played significant role in bladder functional behavior, and suggested that elastin/collagen ratio might be the key factor in determining the compliance of bladder wall.

Myostatin (MSTN), a member of the TGF-beta superfamily, was initially identified as a primary negative regulator of embryonic and postnatal muscle development. The MSTN gene is highly conserved among different species. Mutation of the MSTN gene results in a dramatic increase in skeletal muscle mass in mice, cattle, and humans. To date, most research has focused on the inhibitory role of MSTN on muscle growth, including the MSTN signaling pathway, the underlying mechanism of MSTN function, and the antagonists of MSTN. In this study, we identified a new property of MSTN. The project had 3 primary aims: (1) to characterize MSTN as a fibrogenesis stimulator; (2) to investigate the relationship between MSTN and TGF-beta1, and (3) to investigate the effect of decorin on MSTN. Our findings demonstrate that MSTN stimulates fibroblast proliferation and induces differentiation of fibroblasts into myofibroblasts in vitro. We also showed that MSTN knockout (MSTN-/-) mice develop significantly less fibrosis and exhibit better muscle regeneration than wild-type mice 2 weeks after gastrocnemius muscle (GM) laceration in vivo. In addition, we showed that TGF-beta1 stimulates MSTN expression in C2C12 myoblasts and, conversely, that MSTN stimulates the secretion of TGF-beta1 by C2C12 myoblasts in vitro. In vivo, MSTN injected into the GM stimulates myofibers to transiently co-express MSTN and TGF-beta1. Moreover, TGF-beta1 and MSTN colocalized in the necrotic myofibers shortly after GM laceration. Finally, our results showed that decorin, a natural inhibitor of TGF-beta1, blocks the effects of MSTN. After co-incubating cells with decorin and MSTN, we found that decorin reversed the stimulatory effect that MSTN had on preplate 1 (PP1) fibroblasts and blocked the inhibitory effect that MSTN had on myogenic cells. In vivo, the expression levels of decorin in regenerating muscle are related to MSTN levels. Immunohistochemistry revealed higher decorin expression in MSTN-/- regenerating muscle than in wild-type muscle. Our results suggest that the role of MSTN in injured skeletal muscle is more complex than initially defined: MSTN inhibits muscle growth. MSTN helps to regulate both extracellular matrix deposition in injured muscle and myogenesis. These findings have afforded us a better understanding of the role of MSTN in skeletal muscle healing and indicated that MSTN could be a viable pharmacologic target for antifibrogenesis therapy.