Intralesional curettage versus Prosthetic replacement,which approach is suitable for bone tumors ?------ A finite element analysis of limb salvage simulation in biomechanics

Background : To compare mechanical properties of femoral tumor treatments so that better operative strategy for limb tumors surgery is optimized. Method s : 14 femoral CT images were randomly selected to rebuild 3D models by MIMICS. They were then executed by reverse engineering softwares for simulative modes. Mode #1: Intralesional curettage with cement filled plus fixator; Mode #2: Distal femur resection with tumorous prosthesis replaced. Finally, the mechanical aspects such as stress and displacement were compared by finite element analysis. Result s : Analyzed by AnSys, the observation indexes were measured as follows: for displacement of femurs , d =1.4762 ( < a=3.9042 < c=3.9845 < b=4.1159) in mm is the most staple of all models; for displacement of implants (fixators or prostheses), it’s similar to the behavior of femurs and with no significant difference; for stresses of femurs, no significant difference was found among all models ; the stresses of implants (fixations and prostheses) were observed as d=39.6334< a=58.6206 < c=61.8150 < b=62.6626in MPa correspondently, which is the least; for stresses of the general system, the average of peak values for integrated devices of all models are : d=40.8072 (< a=58.6206 < c=61.7831< b=62.6626) in MPa , which is also the least. As a final result, both maximum values for displacement and stress of mode 2 are lower than those of mode 1. Conclusion : Our finite element analysis of limb salvage simulation in biomechanics proved that , to treat distal femoral bone tumors, p rosthetic replacement is more efficient than intralesional curettage.

final result, both maximum values for displacement and stress of mode 2 are lower than those of mode 1. Conclusion : Our finite element analysis of limb salvage simulation in biomechanics proved that , to treat distal femoral bone tumors, p rosthetic replacement is more efficient than intralesional curettage.

Background
Distal femur is the most common site for malignant bony tumors. Many surgeons treat tumors of extensive bone destruction with wide resection followed by prosthetic replacement, while treat small tumors by intralesional curettage. Over the past decades, 3 these limb salvage methods have evolved and substituted for amputation as the safe mainstay in most surgical cases. However, the best option for limb tumors and the risk factors for recurrence remain controversial. The aim of this paper is to compare the procedures of "Intralesional curettage" with "Prosthetic replacement" [1,2] , their clinical effects on implants will be numerically investigated by Finite Element Method to make suitable decision quantitatively. Finite Element Analysis (FEA) is executed to examine their biomechanical performance by simulation, e.g. identifying displacement throughout structures, and predicting the magnitude of maximal stress [3] . This present investigation was conducted to 2 modes based on comparison among 4 models for optimization during preoperative planning according to the physics emulation, and the insight of computerized findings will enable easier limb-salvage potentially.
In term of malignant bone tumors surgery, we hypothesized that computer aided methods can improve the accuracy of dissection compared to traditional freehand technique, allow precise planning of surgical excision, and develop allograft-prosthetic composite fitting in removal area. Thus, mathematical approximation must be carried out estimating mechanical behavior in order to design surgical resection prior to actual operation [3] . In our experiment, the disease areas were reconstructed three-dimensionally (3D), the cutting ranges were identified, the surgical templates were computer-aided maneuvered, the prosthesis was individualized, tumor resection and reconstruction were emulated [4] .
The primary process includes: by using trimming templates, preoperatively obtained allogeneic bones are trimmed into 3D shapes matching the bony defects after tumor excision; they are then fixed to metal prostheses with screws, bony cement or ingrowth to form individualized prostheses, and finally implanted into the defect areas [5] .

Procedure
Model RebuiltFemoral CT scans of 14 normal volunteers randomly chosen from our hospital during 2019 in *.dicom format were input to medical image processing software called Mimics. After executed by modules of threshold segmentation, erasing, region growing and editing, etc., a 3D model of full femoral was reconstructed as Fig. 1 shows [6] , and was then imported to Geomagic Studio, where osteotomy was virtually cut at 150 mm away from intercondylar process of the femur as Fig. 2 shows. A curved surface of None Uniform Ration B-Spine (NURBS) kyrtograph was fitted after the procedure of "Detection of curvature→Demotion contour line→Structural surface→Structural grid→Fitting curve" was carried out. This NURBS would be input to UG to form a solid model.

Components Implemented
A prosthesis was drawn in the 3D mapping software, ProE, with a stipe length of 150 mm, and output in *.iges format. It was then imported into Hypermesh with the aforementioned solid model, where a tumorous prosthesis was obtained by Boolean Operation. All the models of screws, locking plates, bony cortex strips were mapped in ProE: The diameter of a bony cortex strip is 3 mm, of a screw is 5 mm, with length adjusted to 80 mm based on practical need; the bony cement to be filled in the femoral model was zoomed to 4 pixels, fixed to a suitable shape, and sutured into solid models of mixed cement that is composed 5 of Decalcified Bone Matrix (DBM) and pure Bone Cement (BC) with a proportion of (DBM: BC=) 7.5:2.5.

Models Assembled
The aforementioned components such as femur cement cortex screw plate and prosthesis were imported into a FEM pre-processing software Hypermesh, and were assembled to 4 kinds of models:

Mesh plot & Material assignment
Each part of the models was meshed individually to combine implant entity as Fig. 3-2 shows. They were then transferred back to Mimics to assign material properties which were assumed to follow the empirical formula provided by Mimics: 6 Density=131+1.067×HU, E-Modulus=0.004× density 2.01 . The poisson's ratio was set as 0.3 for femur [7] . After properties and materials sett, the data and models are demonstrated as Table 1 & Fig. 3-3.

Load configuration & Boundary constraint
The above mentioned models were reimported into Hypermesh to set boundary conditions for simulation. There, a pressure of 600 N was applied straight down on the femoral head of an adult weighted 70 kg with one-leg-standing [8] . The degree of each node in xyz direction was constrained to 0. The options of solution were defined, and the finite element models (FEM) were exported in *.cdb format as Fig. 4 shows.

FEM solution
This generated FEM is to be exported onto AnSys for solution of computation, the observation index to be measured includes (1)

Displacement of femur
The displacement distribution of the femurs in these 4 models were observed by reviewing

Displacement of implant
By reviewing the implants' displacement in Fig. 6, we found that the directions of all displacements were also downward vertically, their peak values were focused on the top of the implants, and decrease gradually, concentrically and distally [9] . The sense of displacement for implants including fixators and prosthesis is equivalent to that of femurs.
P>0.05 suggests no significant difference while Kruskal-Wallis rank-sum was examined, and therefore, the mechanical stability of implants is similar to that of femurs.

Stress of femur
As depicted by Fig 7, the contact stresses of all femoral models are concentrated around femoral head and neck. For most "curettage + cement" models, stresses peak values locate at the femoral shaft where implant is fixed by screws; for the "resection + prosthesis" model, peak value locates at the site where prosthesis is connected to the femur [10] . The von mises peak average for each femur is a=29.6990 b=32.5739 c=32.4723 d=25.2321 unit: MPa respectively. The single factor analysis of variance shows P>0.05 suggesting no difference in significance among all models.

Stress of implant
The stress of implants including fixations and prosthesis were observed by reviewing Fig.   8, we noticed that: for models of "curettage + cement + fixator", most of the fixation stresses distribute and peak at upper screws and middle plates, while less at lower screws 8 and plate's ends; for the model of "resection + prosthesis", stress distributes at prosthesis's stem and peaks at upper stem or prosthesis's base connecting to stem. The "single factor analysis of variance" was performed resulting in Table 2

Stress of cement
The stress of bony cement were observed by reviewing Fig. 9, we noticed that, for models of a b and c, most of the cement were well distributed, and peak at its upper side connecting to femur, or its middle region fixed to screws. The Kruskal-Wallis shows P<0.05 suggesting a significant difference. By comparison among these 3 models two by two, is that, the elastic modulus of mixed bony cement (composes of DBM BC = 7.5 2.5) is less than that of pure cement, even almost reach that of cancellous substance, and its peak stress is less than the other models consequently.

Stress of general system
As depicted by Fig 10,

Discussion
This experiment carried out computer aided modeling for surgical management of bony tumors, provided accurate tumor removal with salvage of unaffected bone maximally and precise endoprosthesis reconstruction. Through digital simulation, surgical implants were designed freely adjusting the residual bone. Additionally, the allogeneic bones were trimmed to 3D shapes matching the bone defect based on auxiliary template reducing margins, and they were then integrated to form custom fixtures [11] . On the other hand, by simulating installation, many potential problems could be identified preoperatively, thus operative time and complications must be reduced, mechanical recovery and healing speed would be accelerated.
Displacement implies the stability of a surgical design and the degree of anti-compression for an implant. Observed from femoral data, sub-mode d presented the least peak value, indicates the approach of "Resection + Prosthesis" optimizes stabilization for distal femur tumor rather than "Curettage + fixation", because "the more displace, the more mobilize". However, implants of the 4 models behaved nothing different in displacement, probably because the hardness of fixture and prostheses were not impacted by the load applied. Stress implies the risk of pathologic fracture and implant reversion [12] . In terms of the stresses on femurs, both distribution and peak values of these 4 models were similar statistically. In terms of the stresses upon implants, sub-mode d displayed the least peak value in average, confers that "Resection + Prosthesis" design has a higher resistance from torsional forces and a lower risk of peri-prosthetic breakage rather than the other 3 patterns of "Curettage + Fixator", because, according to the biomechanics principle of clinical relevance for stress, if the same load of force is applied, the higher the level of stress, and the easier it is for the prosthesis or implant to break [ 13] . In terms of the stresses upon integrated system, sub-mode d also behaved more optimal than a, b or c mechanically. Among a, b and c, mixed bony cement (composes of DBM BC = 7.5 2.5) was checked to be biomechanically better than other cement a or c, probably because its elastic modulus is the lowest.
These biomechanical performance generated from pre-surgical designs indicates that "prosthetic replacement" may decrease mechanical failure and revision comparable to "intralesional curettage". Actually, reconstruction with modular customized oncological megaprostheses has become a common procedure nowadays, because wider resection takes advantages of rapid recovery, early weight bearing and lower recurrence, even if smaller resection margins of intralesional curettage allow for preservation of the growth plate and larger areas for endoprosthetic fixation [2,14] . The primary objective of surgical therapy is complete excision of tumors, a secondary goal is conservation of the limb, and the last resort is amputation [6] . Advances in imaging technique, neoadjuvant chemotherapy and surgical methods enable these limb-salvage approaches alternative to amputation with paralleled survival outcomes, even for some cases of benign tumors [1,10] . However, the disadvantage of limb salvage surgery is its complications including neurovascular injuries, local tumor recurrence, deep wound infections, and soft-tissue healing problems [15] . Summarily, limb-salvage of prosthetic replacement not only facilitates malignant tissue removal but also confines local recurrence or metastases [16] , it's playing more and more important role in orthopedic oncology and became a gold standard of treatment.   Von mises for bony cement of mode #l 24 Figure 10 Von mises stress of integrated fixtures