Current Methods of Treatment
       Current treatment options focus on an aesthetically pleasing and immediate solution; however, inducing bone growth provides a long-lasting, integrative solution. Several biological solutions exist for craniofacial defects with area less than 25cm². Patients may receive an autograft, which is considered the gold standard of biologic bone replacement, where a piece of bone harvested from various sources in the patient’s body, usually from the fifth to seventh rib or from the ilium of the hip, is restructured to fill the calvarial defect void.^, While autologously harvested bone decreases the risk of bodily rejection and has high rates of bone fusion, the procedure increases the risk of infection from the graft site, produces multiple incisions which may prolong recovery, and lacks the specific mechanical and geometric properties required to heal the calvarial defect. Similarly, patients may receive grafts from other humans or animals, known respectively as allografts and xenografts. Conversely to the autograft, these grafts have a much higher rate of rejection and more malleable mechanical properties, but do not achieve as high rates of bone fusion. Successful fusion of an allograft or xenograft helps to heal a calvarial defect, but patients may have a lifelong prescription of immunosuppressants to curb the threat of rejection which may lead to greater risks of contracting other illnesses. Synthetic substitutes incorporate a mesh mold covered with polymethyl methacrylate (PMMA) as a personalized mold for the missing skull. Synthetic solutions may have manufactured ideal strength and cosmetic appearance, but lack the integration of bone with the mold and bone growth and have the potential to elicit an immune response for rejection.  Larger injuries may require treatment that can structurally handle the load, namely metal meshes which have many of the same positives and negatives of synthetic treatments^,. 

Figure 1 - An autograft (left) and a synthetic mesh (center) are two typical treatment options for typical calvarial defects (right).

Benefits of Personalized Medicine

     While the current bone healing market has many treatment options available, all have benefits and faults, namely the risk of an immune response for rejection and a lack of personalized fit for optimal mechanical properties for the calvarial defect.

While the aforementioned solutions all provide adequate treatment for patients with calvarial defects, our stakeholders voiced a concern that personalized medicine provides optimal treatment for patients. Dr. David Eckmann of the Department of Bioengineering and Anesthesiology of the University of Pennsylvania advocated for the benefits of personalized medicine stating that patients respond better to personalized treatment. Dr. Ari Brooks of the University of Pennsylvania Department of Clinical Surgery echoed Dr. Eckmann’s sentiments about personalized medicine, and stated that using cheaper materials such as polycaprolactone (PCL) to develop the scaffold will make the treatment less expensive without taking away from the benefits. Personalized medicine and low cost treatment extend well beyond the scope of this course, and ideally, our project will provide better treatment for a lower cost to people around the globe with calvarial defects.

Evaluation Methods

Our review of the current treatment market and the advice of our stakeholders helped our team identify the problem with current treatment: no treatment combines personalized medicine with autologous bone growth and decreases the risk of infection and rejection. Current bone graft methods have insufficient mechanical and biochemical properties, pain at both the implementation and derived sites, rejection rates in up to 30% of procedures, rates of infection up to 10% in patients, and improper geometry to match that of the calvarium.^,,,, Metal and PMMA-based implants have similar issues in terms of infection, mechanical strength, degradation, and potential multiple procedures. The current treatments, while effective to a certain degree, do not provide total patient care with a personalized approach and bone regeneration.

Radiologists will administer some form of imaging to diagnose a calvarial defect properly usually with either magnetic resonance imaging (MRI) or computed tomography (CT). CT stands as the most reliable imaging technique because of the high contrast between bone and native tissue. Our patient treatment process will begin with creating multiple transverse two-dimensional images of the patient’s skull which compile into a three-dimensional representation of the patient’s skull. The radiologist obtains the CT files as a stack of DICOM images, which the user will input into a MatLab algorithm. Our algorithm will use a slice-by-slice reconstruction software to re-create a three-dimensional image of calvarial defect. A 3D printer will take the reconstruction information and will print the respective porous PCL-based scaffold that will optimally fit into the defected skull. The clinician will implant pre-osteoblast and growth factor loaded scaffold back into the native skull to stimulate bone regrowth within the scaffold and native skull. The overall process will take time, but the patient will have native bone regeneration into a complete skull.

Our solution creates a uniform patient-treatment pipeline that produces personalized outputs. Native calvarial bone regrowth is the ultimate end goal of this project, but current advances in tissue engineering have already proven the possibility of regeneration of parietal bone in rat models. The major innovation in this solution is through the creation of a reconstruction pathway that simply requires an input of 3D-CT images and produces an output of a personalized scaffold that stimulates the aforementioned bone regeneration with a manufactured drug release profile. This well-defined pathway will allow for greater patient outcome and less variability in treatment of patients.

Problem Statement

       Over 100,000 patients receive a cranioplasty each year, and many of these patients could benefit from a more robust treatment method. Therefore, we propose a novel, personalized treatment regimen for patients with craniofacial injuries specifically of the calvarial bone. Our project consists of the two following aims: developing a personalized scaffold to place in the calvarial defect and optimizing scaffold drug delivery of growth factors and biologics for sustained release profiles and eventual growth of native bone tissue within the defect. This treatment method will produce a streamlined, personalized process for native bone growth in patients with calvarial defects.

Solution

        The above image shows the overall pipeline of our project. Initially, a patient with a craniofacial defect will undergo CT or MRI imaging so the attending physician can understand the physical extent of the injury in terms of general size, location, and shape. The 3D file of the traced skull obtained from the imaging will export to MATLAB as a stack of DICOM images. A MatLab algorithm will determine the appropriate volume for the calvarial defect from the input images, and produce an output containing the spatial analysis. Afterwards, .STL file containing the 3D stack of images will output to a CAD workstation for 3D printing. The PCL/MSN scaffold will print according to the dimensions determined from the reconstruction portion.

Biologically, the scaffold with optimal porosity and mechanical strength will print for the calvarial defect. The scaffold will bear the several growth factors and pre-osteoblast mesenchymal stem cells (MSCs) loaded in its pores. After biological seeding, a quick surgical procedure will implant the scaffold in the calvarial defect to reduce risk of infection. The scaffold will promote native bone tissue regrowth within the skull while the PCL component degrades and engages the manufactured release profile of the growth factors.

Deliverables

Remaining Challenges

Objectives & Approach Overview

Our review of the current treatment market and the advice of our stakeholders helped our team identify the problem with current treatment: no treatment combines personalized medicine with autologous bone growth and decreases the risk of infection and rejection. Current bone graft methods have insufficient mechanical and biochemical properties, pain at both the implementation and derived sites, rejection rates in up to 30% of procedures, rates of infection up to 10% in patients, and improper geometry to match that of the calvarium.^,,,, Metal and PMMA-based implants have similar issues in terms of infection, mechanical strength, degradation, and potential multiple procedures. The current treatments, while effective to a certain degree, do not provide total patient care with a personalized approach and bone regeneration.

Radiologists will administer some form of imaging to diagnose a calvarial defect properly usually with either magnetic resonance imaging (MRI) or computed tomography (CT). CT stands as the most reliable imaging technique because of the high contrast between bone and native tissue. Our patient treatment process will begin with creating multiple transverse two-dimensional images of the patient’s skull which compile into a three-dimensional representation of the patient’s skull. The radiologist obtains the CT files as a stack of DICOM images, which the user will input into a MatLab algorithm. Our algorithm will use a slice-by-slice reconstruction software to re-create a three-dimensional image of calvarial defect. A 3D printer will take the reconstruction information and will print the respective porous PCL-based scaffold that will optimally fit into the defected skull. The clinician will implant pre-osteoblast and growth factor loaded scaffold back into the native skull to stimulate bone regrowth within the scaffold and native skull. The overall process will take time, but the patient will have native bone regeneration into a complete skull.

Our solution creates a uniform patient-treatment pipeline that produces personalized outputs. Native calvarial bone regrowth is the ultimate end goal of this project, but current advances in tissue engineering have already proven the possibility of regeneration of parietal bone in rat models. The major innovation in this solution is through the creation of a reconstruction pathway that simply requires an input of 3D-CT images and produces an output of a personalized scaffold that stimulates the aforementioned bone regeneration with a manufactured drug release profile. This well-defined pathway will allow for greater patient outcome and less variability in treatment of patients.

Evaluation Methods

Our review of the current treatment market and the advice of our stakeholders helped our team identify the problem with current treatment: no treatment combines personalized medicine with autologous bone growth and decreases the risk of infection and rejection. Current bone graft methods have insufficient mechanical and biochemical properties, pain at both the implementation and derived sites, rejection rates in up to 30% of procedures, rates of infection up to 10% in patients, and improper geometry to match that of the calvarium.^,,,, Metal and PMMA-based implants have similar issues in terms of infection, mechanical strength, degradation, and potential multiple procedures. The current treatments, while effective to a certain degree, do not provide total patient care with a personalized approach and bone regeneration.

Radiologists will administer some form of imaging to diagnose a calvarial defect properly usually with either magnetic resonance imaging (MRI) or computed tomography (CT). CT stands as the most reliable imaging technique because of the high contrast between bone and native tissue. Our patient treatment process will begin with creating multiple transverse two-dimensional images of the patient’s skull which compile into a three-dimensional representation of the patient’s skull. The radiologist obtains the CT files as a stack of DICOM images, which the user will input into a MatLab algorithm. Our algorithm will use a slice-by-slice reconstruction software to re-create a three-dimensional image of calvarial defect. A 3D printer will take the reconstruction information and will print the respective porous PCL-based scaffold that will optimally fit into the defected skull. The clinician will implant pre-osteoblast and growth factor loaded scaffold back into the native skull to stimulate bone regrowth within the scaffold and native skull. The overall process will take time, but the patient will have native bone regeneration into a complete skull.

Our solution creates a uniform patient-treatment pipeline that produces personalized outputs. Native calvarial bone regrowth is the ultimate end goal of this project, but current advances in tissue engineering have already proven the possibility of regeneration of parietal bone in rat models. The major innovation in this solution is through the creation of a reconstruction pathway that simply requires an input of 3D-CT images and produces an output of a personalized scaffold that stimulates the aforementioned bone regeneration with a manufactured drug release profile. This well-defined pathway will allow for greater patient outcome and less variability in treatment of patients.

Plan for Demonstrating Feasibility - Key Deliverables:

Personalized 3D Scaffold: The 3D printed scaffold using the reconstructed file generated from the optimization algorithm with the PCL-MSN solution will serve as our first main deliverable. Demonstrating the feasibility of the construct will start with obtaining publicly-available skull CT images that contain craniofacial defects. The initial algorithm will use the CT images as a test input, and then evaluate the accuracy of the computer-generated defects. Our team will subject the optimized scaffold to physical tests by creating a scaffold for a mock, medical-grade skull. The CT scan analysis will generate the solid, 3D outline of the specific defect., and SolidWorks will use the 3D outline as an input where the hollow interior will be replaced with repeated units of the ideal cross-sectional dimensions to maximize porosity without sacrificing mechanical strength. Upon examining the resulting design by overlaying the model into the original CT stack to ensure proper fit, the construct will be printed using the standard MakerBot. The Department of Mechanical Engineering and Applied Mechanics has available printers that like the ProJet 6000HD, which are able to print layer-by-layer with an accuracy of 0.025-0.05mm. As it is affordable and accessible, one of these high-resolution printers will be utilized to guarantee scaffold resolution of less than 1mm. Once printed, the scaffold will be tested for optimal release kinetics, mechanical properties, and overall fit inside the model skull.

Optimal Mechanical Properties: During the algorithm optimization phase, mechanical testing of the PCL-MSN polymer will be occurring simultaneously. The ideal deliverable through this testing will be that the scaffold itself is able to meet the minimal acceptable mechanical strength profiles of native bone and other clinically-approved scaffolds. First, Dynamic Mechanical Analysis (DMA) will be conducted with an Instron machine to generate a stress-strain curve in addition to modeling creep and stress-relaxation - all of which will provide information about the elastic modulus of the scaffold. The Instron machine will also be used to test the Ultimate Compressive Strength (UCS) of the material. We will test different compositions depending on PCL-MSN ratios and different levels of porosity to ensure the device is able to meet the desired mechanical thresholds. To measure these different levels of porosity, we will utilize a Scanning Electron Microscopy (SEM). This technique generates high resolution images by focusing electron particles at the outer surface of the scaffold. With these images, the size and percent porosity of the scaffold can be calculated and verified to better optimize the relationship between mechanical strength and porosity.

Chemical Testing: The first drug release profiles that will be observed are the antibiotics vancomycin and moxifloxacin, which are commonly used to treat bacterial infections in bones and joints. Because the scaffold are intended for clinical applications to treat cranial defects while overcoming the major drawbacks of current solutions such as high rates of infection, a significant overarching goal is to characterize antibody release profiles as accurately as possible. These drugs will be loaded within the xerogel of the construct for a slow-release profile over the course of 10 weeks to model the time period required for bone regeneration and scaffold degradation. Optimal loading profiles for these drugs will also be monitored to ensure the concentration never reaches an unsafe level as this could lead to negative side effects such as nephrotoxicity and ototoxicity. In-vitro studies using MSCs seeded on the 3D printed material will be conducted to ensure these proper release profiles by assessing cell viability, cell proliferation, and increased levels of bone regeneration as described in “Biological Testing”. Immunosuppressant drugs, such as cyclosporine A, will also be analyzed in a similar manner to limit initial immune response to the scaffold. Grafts and synthetic scaffolds often fail due to the high rates of rejection, so implanting the scaffold with these immunosuppressants will lead to greater mechanical strength of the scaffold and more sustained bone growth.

Biological Testing: MC3T3 cells are mouse osteoblast cells, which will be cultured onto the scaffold through two methods: (1) seeding after the PCL scaffold is printed and placed in a xerogel solution, or (2) direct seeding through bioprinting with cells in the same solution as the PCL:MSN, which will then be fed through the 3D printer. The latter mode of embedding cells has been explored by many labs, including that of Dr. Jason Burdick of the Department of Bioengineering. In order to ensure homogeneity, high retention rate, and direct integration within the macroenvironment, cells can be 3D printed in a cell culture hood. However, indirect seeding is beneficial in that cells naturally bind to PCL due to its intrinsic properties, and the viability would be higher than rough injection due to the large quantity of cells killed during printing. Cytotoxicity is a significant factor in determining if the 3D printed scaffolds are a viable alternative to the currently used materials - many metal scaffolds eventually fail. Experiments such as the LDH and MTS assays can be utilized to understand the viability of the cells and the cytotoxicity of the environment. For MTS assays, it is known that proliferating and viable cells reduce the concentration of MTS tetrazolium when exposed, so a high level of colorized MTS indicates high degrees of cytotoxicity. Therefore, these results are the most important in proving the feasibility of the project, as negative results would necessitate the reevaluation of every aspect of the scaffold to determine the limiting factor.

Current Progress:

Reconstruction: In terms of the reconstruction efforts so far, we have been able to make some major improvements that will allow for less required work next semester. First, we were able to access publicly-available CT images online and download several sets of .dicom images of healthy skulls. With this data, we have been able to successfully read-in 3D CT and separate images on a slice-by-slice basis for reconstruction. Within these individual slices, we have been able to make significant progress in our reconstruction efforts. As seen in the image, the progress to this point can be visualized in six steps. (1) shows our ability to read in CT data on a slice-by-slice basis for reconstruction. Step (2) shows a conversion of the CT image to binary, using the most sensitive possible threshold, which was advised by Dr. Ari Brooks. Image panes (3) and (4) show our capabilities to create outlines of the skull depending on different borders and sides of the skull, which will be helpful when localizing defects in 2D. (5) represents current geometrical modelling that we have employed by using code to fit an ellipse to the inner and outer borders of the skull through a least-squares regression. Finally, (6) shows that, with these geometrical models, we can determine certain properties of the skull, such as the labeled axis of rotation.

Biologics: Biologically, we accomplished one of the important principle synthesis reactions necessary for our scaffold’s manufactured drug release profile: fabrication of blank mesoporous silica nanoparticles. We mimicked a method developed by Sanjib Bhattacharyya, Henson Wang, and Paul Ducheyne as detailed in “Polymer-Coated Mesoporous Silica Nanoparticles for the Controlled Release of Macromolecules” in Acta Biomaterialia. Citation/footnote (I have paper). First, C₁₈TAB (octadecyltrimethylammoniumbromide) was dissolved in distilled water at 75℃, and then mixed with 2M NaOH (sodium hydroxide) to make a basic solution with pH 12. The main molecule, TEOS (tetraethylorthosilicate) was dissolved in with drops of TESPA (3-(triethoxlysilyl) propyl-succinic anhydride), and the solution was left to dry for two days. The dried MSN particles were added to aqueous DMH (dimethylhexadecylamine), stirred, and then heated. The leftover solid was rinsed with methanol first, followed by rinses with ethanol and water. This procedure creates pore-expanded, unloaded MSN particles which can take on a biologic to load its pores  when we design the PCL/MSN solution. The figure Number(MSN stages graphic, obtained from paper) above shows the main stages of the MSN fabrication and loading process with the last step of the PEG-coated MSN as the final product necessary to produce the manufactured drug release profile.

Specifications

Specifications:

Design Goals, Tradeoffs and Constraints:

Mechanical Properties: The personalized implant scaffold will be fabricated with a Young’s Modulus >= 2.86GPa and an ultimate compressive strength (UCS) >=70MPa.  Clinical studies have shown that the ‘gold standard’ procedure for craniofacial reconstruction, the autograft with bone sourced from parietal bone, has exhibited a Young’s modulus ranges from approximately 0.9 to 15.54 GPa and an average UCS of 100MPa^,,,,. Further studies indicate that cranial implants experience additional degenerative forces: internal pressure caused by tensile forces within the skull that range from 80-100kPa, external pressures caused by impacts to the skull, and pressure applied by the skin, which results in both shear and tensile forces. These mechanical properties vary within the calvarial bone depending on several factors (eg. loading speed, location within calvarium bone, skull thickness, age, etc.), but human skulls rarely experience these instantaneous pressure changes of these magnitudes. The scaffold should have the appropriate mechanical properties to withstand these forces and pressures, but patients with the scaffold should limit exposure to excessive external forces on the skull.

Currently, PMMA-based cements (among other synthetic treatment options) have shown that pure PMMA-based cements have a Young’s Modulus of around 2.86GPa and a UCS of approximately 93.0MPa^,. When we reached out to our stakeholders, Dr. Brooks advised our team that the mechanical properties of the scaffold do not need to equal those of autologous bone growth, but must display enough strength so that the scaffold can withstand the surgical procedure for implantation and maintain the structural integrity of filling the calvarial bone.

Porosity: The personalized implanted scaffold will have an average porosity of 60-90%.  Studies show that porosity of 60-90% provides the optimal range for seeding the scaffold with cells, drugs, growth factors, and other biologics. The pores must be interconnected to develop and maintain a vascular system required for continued bone development^,,. We will optimize the ideal percent porosity when we run mechanical testing due to the inverse relationship between porosity and mechanical strength. No major “trade-offs” will be made, as we will, ideally, pick the most porous material that is able to meet the minimal limit of the defined mechanical strength properties. Additionally, the average pore size for all pores within the generated scaffold will fall between 200 to 350µm in diameter, and no pore will have a pore size of 150µm or smaller due to the risks of improper seeding and release of pre-osteoblast cells. Small pores allow the biologics and growth factors to seed initially in the scaffold, but will allow also for vascularization and ossification within the pores to ensure proper  bone growth.

Reconstruction: Critical-sized cranial defects will be reconstructed into a 3D CAD model from a stack of axial CT images. As previously stated, we will initially limit the scope of the project to only the calvarium of the skull, which incorporates two major bones: the parietal bone and the frontal bone. Other bones do not clearly reflect the ovular shape of the skull in axial CT, so reconstruction will require more advanced algorithms. With the reconstruction itself, we will be able to take any patient’s 3D-CT stack and produce a personalized output within a matter of minutes. Specifically, we will develop the capacity to load in any stack of .DICOM files corresponding to 3D-CT images supplied by a radiologist. Individual slices will then be analyzed from the stack, and with this data, an algorithm will be developed to identify and trace the missing section of skull within a single slice. These individually-corrected slices will then be compiled into a 3D stack and exported as a CAD-compatible .STL file.

Reconstruction Accuracy: Within each cross-section, modeled data points will be plotted along the peripherals of the skull geometry (inner and outer borders), with a minimum accuracy of 90%.  By outlining the skull with thousands of data points, we will fit an ellipse, among other geometrical figures, to these data points and estimate the skull geometry to this high accuracy for reliable reconstruction. Although ellipses will likely not be used as the final reconstruction geometry, these data points are fit through a least-squares regression to minimize  the distance from the borders of the skull to the implant.  Within a single cross-section, the plotted data will be scaled to accurately resemble the true dimensions of the patient’s skull.  Data points within the CT-images will not be true dimensions of the patient’s skull, so the algorithm will utilize a scale bar within the CT slices to ensure the scaffold has the correct dimensions when printed.

3D Printing: When compiling the two-dimensional cross sections of the CT slices, the third dimension in the superior direction of the axial plain will have a slice thickness of <1.5mm. Although CT scans of the smallest clinically feasible slice thickness of the skull, 0.625mm are ideal, studies have shown that CT slice thickness <1.5mm showed no significant difference in quantitative properties (eg. volume, surface area) of the skull. Images acquired at a thickness greater than 1.5mm have demonstrated significantly different reconstruction values, so they will be avoided for the purposes of clinical feasibility.  The average height of 3D-printed PLA is 400 µm, so we anticipate the resulting constructs will fall within a standard deviation of 400 µm due to the major limiting factor of low resolution of the CT scanner itself.

Constraints: For the scope of this project, the primary constraint is the budget. Due to the pre-set budget of $600,  the magnitude and selection of technology, material, and testing are limited. Specifically, it is particularly expensive to gain access to imaging modalities such as CT or MRI within the School of Medicine and Singh Nanotechnology, to buy specific growth factors. Secondly, there are significant time constraints for the entirety of the project, as we will not be able to conduct many of the assays we hope to prove feasibility. Ideally, the implant will go through the inchoate prototype stage followed by the first round of dry evaluation and in-vitro testing. However, the fast-approaching deadline limits the project to the prototyping and, if even possible, simpler in-vitro testing. This rudimentary testing will allow us to evaluate the proof-of-concept and feasibility; however, the deadline inhibits the project from proceeding to advanced in-vitro, ex-vivo, or in-vivo testing stage, where the majority of the biological feasibility work would occur. Finally, limited meeting times between group members and our advisors is a constraint that we face. We would like to be able to brainstorm with our advisors and each other on a more frequent basis than what scheduling has so-far allowed. 

Regulatory Pathway: Our advisor has several years of experience dealing with United States and European regulatory committees in the field of medical device production, so we are confident that the physical biomaterials used in this project would qualify for the 510(k) regulatory pathway within the FDA as a Class-II medical device. So long as the material used is deemed “substantially equivalent” to other clinically-approved products such as PMMA-based scaffolds, then the PCL/Xerogel-based substance will likely be approved with various biologics and drugs. As discussed previously, there exists a number of implantable scaffolds for the application of craniofacial reconstruction, although few are largely-composed of 3D-printed biological material. Thus, there is a possibility that the overall treatment pathway will require Pre-Market Approval (PMA) from the FDA due to the invasiveness during implantation and potential direct human injury with its use. That is, the scaffold itself would likely only required 510(k) approval due to similarities with other devices currently available. However, the overall treatment pathway which utilizes the scaffold will require PMA. Additionally, the FDA is generally known to be wary of algorithm-based clinical treatment due to the lack of human overview of the device. Therefore, in producing this device, it is likely that large-scale randomized, controlled trials will need to be run in comparing the efficacy of the proposed treatment pipeline compared to current clinical standards.

---

[1] https://www.fda.gov/downloads/RegulatoryInformation/Guidances/ucm126054.pdf

[2]https://www.fda.gov/downloads/BiologicsBloodVaccines/GuidanceComplianceRegulatoryInformation/Guidances/Tissue/ucm062592.pdf

[3] https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3562252/