Mechanical behavior of implant-supported full-arch prostheses in different locations in the maxilla: 3D-FEA and strain gauge analysis

The maxillary bone restriction can limit the implants position to support a full-arch prosthesis. Objective: Therefore, this study evaluated the biomechanical behavior of a full-arch prosthesis supported by six implants in different configurations: group A (implants inserted in the region of canines, first premolars and second molars), group B (implants inserted in the region of first premolar, first molar and second molar) and group C (implants in second premolar, first premolar and second molar). Material and Methods: The models were analyzed by the finite element method validated by strain gauge. Three types of loads were applied: in the central incisors, first premolars and second molars, obtaining results of von-Mises stress peaks and microstrain. All registered results reported higher stress concentration in the prosthesis of all groups, with group C presenting higher values in all structures when compared to A and B groups. The highest mean microstrain was also observed in group C (288.8 ± 225.2 μ ε/ μ ε ), however, there was no statistically significant difference between the evaluated groups. In both groups, regardless of the magnitude and direction of the load, the maximum von-Mises stresses recorded for implants and prosthesis displacements were lower in group A. Conclusion: It was concluded that an equidistant distribution of implants favors biomechanical behavior of full-arch prostheses supported by implants; and the placement of posterior implants seems to be a viable alternative to rehabilitate totally edentulous individuals.


INTRODUCTION
Implants with external hexagon-type connections presents biomechanical characteristics with acceptable immediate performance as well as in the long-term follow-up.For this and other reasons, fixed prostheses on external hexagon implants are a very popular option in the treatment of edentulous patients [1].
However, for rehabilitation with an implant-supported restoration, it is necessary to have adequate bone remnant to support the masticatory loads.After tooth extraction or tooth loss, the alveolar bone undergoes a physiological remodeling process that often limits the amount of bone, especially in the anterior region of the maxilla; dampening the oral rehabilitation in this region [2].Thus, there are two major clinical procedures that can provide the necessary anchorage in these areas: bone grafting or the use of long implants in the posterior region with anchorage in other portions of the bone tissue [3].It is important to emphasize that the use of long implants with zygomatic anchorage, in addition to being invasive and requiring hospitalization, usually has a high cost to the patient [4].
A clinical option that overcomes these limitations would be the implant placement in the posterior region of the maxilla.However, this treatment option can present complications in the long-term due to the lack of standardization of it and mechanical studies supporting its indication [5].The two major factors of implant failure are: peri-implantitis and occlusal overload.Both factors can act together or independently and cause peri-implant marginal bone loss that, in advanced cases, can lead to implant loss [6,7].
When the chewing loading mechanical stimuli are within physiological limits, they will result in the maintenance of the bone level, however, when the stimuli exceed the physiological limits, the consequence is a bone loss caused by the disorganization of the remodeling process [8].
Therefore, to avoid the marginal bone loss, it is of great importance to know how the masticatory loads can modify the biomechanical behavior of implant prosthesis [9], since the condition of the marginal bone of an implant in function is influenced by transmitted occlusal forces to him [10].
Axial loads transmit stresses along to the implant axis more homogeneously than oblique loads, being considered more friendly by the peri-implant bone tissues [11,12].However, the positioning of implants and the framework of a prosthesis on implants can influence the distribution of occlusal loads and result in a greater presence of oblique loads, which intensify the magnitude of stresses transferred to the marginal bone [13].In order to improve the understanding of the biomechanical behavior of fixed prostheses on implants, in vitro and in silico studies have been used through bioengineering tools, such as the numerical analysis using the finite element methods [14].
Finite element method allows simulating the possible stresses in the theoretical model.This methodology has the advantage of allowing simulation of various well-controlled conditions, allowing the analysis of the biomechanical behavior of implants in areas of difficult clinical access [15].However, it can give more reliable results when associated with in-vitro methods such as strain gauge analysis.Therefore, this research aimed to evaluate the stress and strain distribution of implantsupported full-arch prostheses with different implant configurations and load conditions.The null hypothesis was that the microstrains would be similar regardless the simulated condition.

3D modelling
T h e e x t e r n a l h e x a g o n i m p l a n t a n d prosthetic screw were created according to the manufacturer's dimensions (Intraoss, Sistemas significativa entre os grupos avaliados.Em todos os grupos, independentemente da magnitude e direção da carga, as tensões máximas de von-Mises registradas para os implantes e deslocamentos de próteses foram menores no grupo A. Conclusão: Concluiu-se que a distribuição de implantes de forma equidistante favorece o desempenho biomecânico das próteses de arco total suportada por implantes; e o posicionamento de implantes posteriores parece ser uma alternativa viável para reabilitar indivíduos densdentados totais.

PALAVRAS-CHAVE
Implantes dentários; Fenômenos biomecânicos; Prótese dentária;, Análise de elementos finitos; Maxila.After that, the computational analysis was performed simulating this reference model in three groups with different configurations of implant placement and transferred to a CAD software for the elaboration of the volumetric model.
The STL file was converted using the Rhinoceros software (version 5.4.2SR8, McNeel North America, Seattle, WA, USA), using reverse engineering tool.The fixed prosthesis on the implants were modeled with the same steps as the maxilla from the STL file generated by the CAD software.After, the 3D model was finally finished as a volumetric model (Figure 1).

Boundary conditions
After modelling, the three-dimensional model was imported into Ansys software (ANSYS 16.0, ANSYS Inc., Houston, TX, USA) in order to carry out a static structural analysis.Material properties were used from software database.The geometries were renamed according to what they are representing, and all structures were considered linear, homogeneous, isotropic and elastic.After checking the contact between the structures, they are considered bonded and the number of faces tangent between two solids were adjusted with similar quantity (Figure 2).
The meshing process has been carried out automatically, in which the software allowed the refinement of the mesh created using tetrahedral elements.For each load (axial and non-axial) an analysis configuration was used.For all configurations, the maxilla was fixed on its lower external surface simulating the support of the model in a plane.Load was defined as vector in the Z-axis direction with 300N force on premolars and molars, 100N on maxillary central incisors.After the simulations, von-Mises stress solutions were conducted for the implants, prosthetic screws and the maxillary model for each load (Figure 3).

In-vitro strain assessment
To validate the FEA model an in-vitro model was design with similar conditions.The surfaces of the model were prepared and cleaned with isopropyl alcohol and electric linear strain gauges (KFG-02-120-C1-11, Excel Sensores Industria e Comercio., Ltd -Taboão da Serra-SP, Brazil) were bonded.Each sensor was positioned close to the implant using a transparent adhesive tape and a cyanoacrylate-based adhesive (Super Bonder Loctite, São Paulo -Brazil).
The model received similar loads following FEA with a 2 mm diameter rounded tip as the load application device [16].Variations in electrical resistance were transformed into microstrain units using an electrical signal conditioning device (Model 5100B Scanner -System 5000 -Instruments Division Measurements Group, Inc. Raleigh, North Carolina -USA, FAPESP proc: 07/53293-4).Data recording was performed using strainsmart software.Electrical cables allowed the connection between the strain gauges and the data acquisition device (Figure 4).It was possible to observe for each image that the hot colors represent the zones with the highest stress concentration (von-Mises Stress), that is, regions under higher stress and for each analyzed structure.

RESULTS
The implant-supported prosthesis was the structure with the highest stress peak for the three groups, which could be observed that in the region of the most mesial implants.However, the maxillary model did not show a high magnitude of stress regardless the different conditions, in addition to the mesial region of the first implant (figures 5-7) (Table I).Therefore, for the oblique loads applications, it was possible to observe a lower concentration of stress and deformation in the structures furthest from the loading regions and higher stresses in the regions closer to the load application point.When the load was ap-plied in a region of the first molar of the prosthesis, the stresses were distributed more homogeneously between the evaluated structures.
Observing the deformation distribution generated in group C, it is possible to notice that the magnitude of the deformation peak is concentrated in the more mesial implants when the load applied to the centrals compared to groups A and B.
For strain gauge analysis, statistical tests were performed using the R-project 3.2.0software (Table II).The significance level established for

Grande MFB et al.
Mechani cal behavi or of i mpl ant-supported ful l -arch prostheses i n di fferent l ocati ons i n the maxi l l a: 3D-FEA and strai n gauge anal ysi s

Grande MFB et al.
Mechanical behavior of implant-supported full-arch prostheses in different locations in the maxilla: 3D-FEA and strain gauge analysis the tests was equal to 5%, which established a 95% confidence interval for the presented results (Table III).Therefore with similar mechanical behavior between in-vitro strain peaks and theoretical stress peaks, the model has been considered validated.
With values of p < 0.05 for the Shapiro-Wilk test, the distribution of deformations (Strain) was considered non-normal.Kruskal-Wallis statistical test was applied to assess the relationship between each model and its strains.The results obtained showed no statistical difference (p = 0.902) for the evaluated groups when considering the average strain per model (Table III).
For each load application point, the Kruskal-Wallis test and Dunn's multiple comparison test were used.It was possible to observe that there was no statistical difference for the different loading conditions (X2 = 2.486, df = 5, p = 0.778).

DISCUSSION
This study evaluated the influence on external hexagon implants in three different designs of multiple implant-supported prostheses in edentulous maxilla.The null hypothesis was partially accepted since the microstrain were similar between the groups.However, it was also possible to observe that the greater the lever arm in the anterior region, the greater the magnitude of the stresses calculated for the implants and structures of the prosthetic system.
According to the literature [17], in silico methods such as finite element analysis, can be used to measure bone behavior.The use of such 3D method requires prior knowledge of bone volume and mechanical properties.However, the bone tissue is not homogeneous and its physical properties vary greatly according to species, age, sex, type of bone (e.g., femoral, mandibular, cortical, cancellous) and even according to location of the bone from which the sample was taken [18,19].
Therefore, studies with human bone have a complexity and heterogeneity and, for ethical reasons, often delay the development of clinical trials.Thus, through an in vitro study, a previous study evaluated the elastic modulus of an experimental polyurethane isotropic model, by means of stress tests and compared the results with those reported in the literature with bone [19,20].According to them, the use of the polyurethane model in place of bone in in vitro biomechanical studies is a validated method.Based on that, the present study used the in vitro polyurethane model and simulated the numerical model with the properties of this isotropic model.
In the present results, the disposition of the implants to distal in the edentulous maxilla caused a higher stress concentration for the analyzed structures, in accordance with previous studies that reported a similar behavior, when the distance between the implants increased [21][22][23].
A previous study [24] reported that implants installed respecting the minimum space between each implant (3 mm) demonstrated a better mechanical response of the bone tissue to masticatory loads.In this way, the use of six adjacent implants in the posterior region (Group C) allowed an acceptable mechanical behavior of the peri-implant bone under load.Therefore, the results obtained for the different implant positions clearly indicated that the stress was lower in model A compared to other models B and C.
The stress in the implant and in the framework increased as the implants were installed distally to the molar region.Therefore, the more distal the configuration of the implants, the greater the distance between the anterior abutments of the implant.When the load was on the central incisors, the amount of stress concentration were visibly higher compared to the similar loading condition in model A. This finding is consistent with the study of previous authors [25], who found that cantilever loading had a large effect on stress concentration.Increased stress can increase the risk of treatment failure [12].The results obtained in the present study also indicated that anterior cantilever can negatively affect the implant stress magnitude.
Although the positions of implants B and C can be considered almost similar, the loading on the central incisors increased the stress on the mesial implant in group C compared to group B, showing that these designs present different mechanical response.
Another novelty of the present study in relation to previous reports in the literature was the verification of stress, considering a cantilever prosthesis in the anterior maxillary implants.Although this factor has been evaluated in previous studies [12,[21][22][23][24][25][26], there is lack of information when considering different implant positions.The literature [21][22][23][24][25][26][27] usually reports the maximum forces capable of generating mechanical problems in the bone-implant interface, however, there are no reports showing the cantilever in anterior region.
It is noteworthy that this study was subject to some limitations, as the loading condition was simplified [28][29][30].Axial loads transmit stresses along the axis of the implant in a more homogeneous way, being better accepted by the peri-implant bone tissues [31,32].However, the positioning of the implants and the framework of a prosthesis on implants can influence the distribution of occlusal loads and result in a greater amount of oblique loads, which intensify the magnitude of the stresses transferred to the marginal bone [33].In order to improve the understanding of the biomechanical behavior of fixed prostheses on implants, in vitro and in silico studies have gained notoriety through bioengineering tools, such as, for example, analyzes by FEA and strain gauge [30][31][32][33].
Future studies should apply different bone densities, anatomical structures such as the maxillary sinus and diversify the size and angulations of implants for this situation.Furthermore, although the experimental models were rigorously prepared and experienced dentists were involved in all procedures, the bone model is still an isotropic structure that is limited in terms of mechanical response.Therefore, it is not recommended to extrapolate these results to implant-supported prostheses in clinical situations, and further studies should be carried out to assess the effect of traction forces on the biomechanics of these prostheses.However, despite all the limitations, the present results can be used to guide further in-vitro studies and to elucidate how the implants distribution and loading condition can modify the mechanical behavior of the full-arch prothesis on implants.

CONCLUSIONS
According with the obtained results and model validation it is possible to conclude that the implants inserted in the region of canines, first premolars and second molars showed the most promising mechanical behavior, while the distal implants placement showed the highest stress.However, the positioning of implants in the posterior region of the maxilla seems to be a viable alternative to rehabilitate edentulous maxilla.

Figure 1 -
Figure 1 -(A) Model generated based on the external geometry of the pre-existing model; (B) Implant-supported prosthesis designed in CAD software; (C) Model was transformed into a volumetric solid; (D) Pre-existing physical model that served as the basis for the digital archives and for the in-vitro strain measurement.

Figure 2 -
Figure 2 -Model A: implants inserted in the region of canines, first premolars and second molars.Model B: implants inserted in the region of the first premolar, first molar and second molar.Model C: implants inserted in the region of second premolar, first premolar and second molar.
Grande MFB et al.Mechani cal behavi or of i mpl ant-supported ful l -arch prostheses i n di fferent l ocati ons i n the maxi l l a: 3D-FEA and strai n gauge anal ysi sGrande MFB et al.Mechanical behavior of implant-supported full-arch prostheses in different locations in the maxilla: 3D-FEA and strain gauge analysis

Figure 3 -Figure 4 -
Figure 3 -Loading simulation: (A) Oblique loading in central incisors; (B) simulation of axial loading in first premolars; (C) axial loading on second molars; (D) model fixation for load application and, (E) perspective view of the different loading conditions.

Figure 5 -
Figure 5 -Von-Mises Stress (MPa) results in the bone tissue for loading in central incisors (A-C), in the first premolars (D-F) and upper molars (G-I) according to the three different models.

Figure 6 -
Figure 6 -Von-Mises Stress (MPa) results in the dental implant for loading in central incisors (A-C), in the first premolars (D-F) and upper molars (G-I) according to the three different models.

Figure 7 -
Figure 7 -Von-Mises Stress (MPa) results in the prosthetic screw for loading in central incisors (A-C), the first premolars (D-F) and upper molars (G-I) according to the three different models.

Table I -
Stress peaks (MPa) calculated according the different models (A, B and C) as well as the loading region for each of the evaluated structure Grande MFB et al.Mechani cal behavi or of i mpl ant-supported ful l -arch prostheses i n di fferent l ocati ons i n the maxi l l a: 3D-FEA and strai n gauge anal ysi sGrande MFB et al.Mechanical behavior of implant-supported full-arch prostheses in different locations in the maxilla: 3D-FEA and strain gauge analysis

Table II -
Strain peaks (microstrain) calculated according the different models (A, B and C) as well as the loading region (Incisors, Pre molars and Molars) for each of strain gauge

Table III .
Kruskal-Wallis table according to each model Mechani cal behavi or of i mpl ant-supported ful l -arch prostheses i n di fferent l ocati ons i n the maxi l l a: 3D-FEA and strai n gauge anal ysi s