Two bovine hydroxyapatites (BHAs), one with granule size of 150 to 200 μm and one with granule size of 300 to 329 μm, and 2 synthetic hydroxyapatites (SHAs), with granule size of 150 and 300 μm, respectively, were compared for effectiveness in repairing circumferential bone defects in dogs. The hydroxyapatites (HAs) were characterized through powder x-ray diffraction (XRD) analysis and scanning electron microscopy (SEM). Three trephined bone defects (5.0 mm wide × 4 mm long) were created in the humeruses of 8 dogs. In a random manner, the defects on each side were treated with either BHA with small granules (BHA[s]), BHA with large granules (BHA[l]), SHA with small granules (SHA[s]), SHA with large granules (SHA[l]), or left to heal unaided (bilateral control). Four dogs were sacrificed after 6 and 12 postoperative weeks, respectively. Ground sections of each defect were submitted to histologic and histomorphometric analysis (percentage of area occupied by bone, bone marrow, and biomaterial). As a rule, the HA granules exhibited direct bone contact, regardless of the origin and the size of the granules. Control sites were related and had an increased amount of connective tissue infiltration. At 12 weeks, BHA(s) exhibited improved bone formation compared with SHA(s) and SHA(l). The SHA(s) delivered reduced amounts of bone compared with the remaining groups (control included). The area of bone measured in BHA(s) sites was significantly higher at 12 weeks than 6 weeks. The XRD revealed the tested HA samples to be highly crystalline, while BHA appeared with rougher surface at SEM analysis. The BHA(s) performed better than the SHA(s) and SHA(l), as assessed by the amount of bone measured in both implantation sites at 12 weeks. The BHA's material characteristic itself rather than granules size accounted for the distinctive biological behavior. The increased roughness of the BHAs' surface, as assessed through SEM, seemed to benefit the osteoconduction process.
The spontaneous alveolar ridge healing that follows teeth extractions frequently leads to bone deformities that result in unsuitable sites for titanium implant placement.1,2 Several granulated biomaterials have been tested as therapeutic adjuvant to obviate the formation of bone defects in postextraction dental sockets as well as for sinus floor augmentation procedures, either used alone or mixed with autogenous bone.3 Among these biomaterials, bovine and synthetic hydroxyapatite (HA)—Ca10(PO4)6(OH)2—granules have gained particular interest over the recent decades because of the multidisciplinary clinical application.2,4,–7
The mechanisms behind tissue response to HAs have not as yet been comprehensively established in the literature. A number of studies have demonstrated that osseointegration and osteoconduction processes are influenced by physical and chemical properties of the material, including granule size, granule morphology, crystallinity and porosity, surface roughness, and ratio of calcium to phosphate (Ca:P) in the composition. Nevertheless, the requirements the best performance for these biomaterials have not been thoroughly addressed.
Oonishi et al,8 evaluated HA with granules of 1 to 3 μm, 10 μm, and 100 to 300 μm in diameter and noticed that a minimal size of 10 μm was necessary to enable a direct contact between bone and the particles. Sun et al,9 studied the effect of different sizes of HA granules (from 0.5 to 841 μm) in osteoblast cultures and reported inhibitory effects for the 0.5 to 3 μm group. Kuroda10 evaluated bone defects filled with HA granules of 100 to 2000 μm and observed improved osteoconductive activity in the group with granules of 100 to 300 μm.
Misiek et al11 compared sharp-edged and rounded HA granules and observed that although a mild inflammatory response was seen at the implant sites with both particle shapes, inflammation resolved faster in sites implanted with rounded granules.
Yang et al12 stated that the dissolution and crystallinity of HAs were related in a negative manner in vitro. The Ca:P ratio in the composition of HAs seems to directly affect crystallinity. Thus, highly crystalline HA—Ca10(PO4)6(OH)2—has a Ca:P molar ratio of 1.67, whereas less crystalline bioceramics, such as tricalcium phosphate and tetracalcium phosphate, are characterized by lower or higher ratios of 1.50 and 2.0, respectively.13
Takeshita et al14 observed that nonporous HA granules grafted into bone defects surrounding titanium implants resulted in fibrous encapsulation during the early healing stages.
Deligianni et al15 used a bone marrow cell-culture model to demonstrate that cell adhesion, proliferation, and detachment strength increased as the roughness of HA increased.
To date, a wide source of HA has been made available for clinical application. Synthetic hydroxyapatites (SHAs) are the most frequently used, but they do not completely match the chemical composition of bone. Bovine hydroxyapatites (BHAs) have recently been used as an interesting alternative to SHA for bone regeneration procedures because of the lower production cost and because they naturally derive directly from the bone itself. In a rare in vivo comparison between SHA and BHA, the former delivered higher elasticity modulus than both BHA and the resident bone after 6 and 26 weeks of implantation in rabbit femur. The BHA equalized the compressive strength of natural bone.16 It is expected that biomaterials with high modulus of elasticity can adversely influence the remodeling of the surrounding bone, as the mechanical stimulus is impaired from evenly dissipating through both sites.17
The distinctive biological and mechanical properties exhibited by these biomaterials have raised the attention of many investigators. While widely studied in the literature, to our knowledge the influence of the granule size of synthetic and bovine bone substitutes has not been addressed by in vivo studies. The purpose of the present study is to compare the performance of BHA and SHA tested in 2 different granule sizes in a bone-healing model in the dog humerus.
Material and Methods
The HAs were supplied by the Department of Industrial Engineering, University of Marmara, Turkey. The BHAs were produced according to the method described by Oktar et al.18 The bovine bone obtained from diaphyseal tibia was irrigated with tap water and soaked in a 1% concentration of an antiseptic solution. Subsequently, the bone was reirrigated and deproteinized in an alkali solution (1% concentration of sodium hypochloride). Thereafter, samples were reirrigated with tap water and calcined at 5°C/min to 850°C for 5 or 6 hours. The matter was sintered at 1200°C for 4 hours, ground in a ball-grinder, and sieved into particle sizes of 150 to 200 μm and 300 to 329 μm. The SHAs were prepared from apatites obtained by hydrolysis and subsequently sintered at 1200°C for 4 hours. The matter was ground in a ball-grinder and sieved into particles sizes of 150 to 300 μm. The HA portions allotted to each bone defect (55 mg) were placed in polymer sterilizing bags and subjected to ethylene oxide sterilization (IPEN—Nuclear Power Research Institute, São Paulo, Brazil).
Eight young male mongrel dogs weighing between 15 and 20 kg were used in this study. The research protocol as described subsequently was approved by the local Animal Research Ethics Committee at the University of São Paulo at Ribeirão Preto.
The animals were sedated with intramuscular 5.0 mg/kg xylazine (Dopaser, Calier Laboratories, Barcelona, Spain) and 20 mg/kg ketamine (Ketamine Agener, National Pharmaceutics Chemistry Union, Embu-Guaçu, Brazil) just before surgery. Afterwards, they were anesthetized with 1 mL/kg of endovenous thiopental sodium (Anental, 20 mg/kg thiopental diluted in 50 mL saline solution, Strides Arcolab Limited, Secunderabad, India).
After the legs were shaved and the skin was disinfected with iodine tincture, an incision was made at the midshaft of the humerus intended for full-flap reflection. Using a trephine (3i, Implant Innovations Ltd, Palm Beach Gardens, Fla) (5.0 mm wide × 4.0 mm long), 3 perforations separated by 10 mm distance were created in the humeral bone (Figure 1) so that every animal had 6 bone defects each.
The defects were divided into 5 groups according to type of treatment: BHA with small granules (BHA[s]) (150 to 200 μm); BHA with large granules (BHA[l]) (300 to 329 μm); SHA with small granules (SHA[s]) (150 μm); SHA with large granules (SHA[l]) (300 μm); and control (sham defect). The treatments were randomly allocated.
Each perforation was filled with 55 mg of granules mixed with 0.1 mL blood to facilitate manipulation and insertion into the defect3 (Figure 2). In all cases the amount of material implanted was sufficient to fill the bone defect.
After the materials were implanted, the flap was repositioned using separate sutures for periosteum and muscular fascia (Vicryl 4–0, Ethicon, Johnson & Johnson, São José Dos Campos, Brazil). The skin was closed with nylon stitches (Nylon Monofilament 5–0, Brasuture, São Sebastião da Grama, Brazil).
At the end of the surgery the dogs received a single dose of intramuscular flunixin meglumine analgesic (Banamine, 1 mL/10 kg, Schering-Plough, Rio de Janeiro, Brazil) and 24 000 IU of penicillin/kg along with 10 mg/kg of streptomycin (Pentabiótico, 0.5 mL/5 kg, Ford Dodge, Campínas, Brazil) antibiotics. The dogs were allowed to consume water and dog chow ad libitum and were checked daily for postoperative complications.
After 6 and 12 postoperative weeks, respectively, 4 animals were sacrificed using a lethal dose of thiopental. One bone block was removed from each humerus, which encompassed the entire region with the 3 defects.
The bone blocks were fixed in 4% formalin for 10 days, dehydrated in increasing alcohol grade concentrations up to 100%, and finally embedded in LR White resin (London Resin Company, Berkshire, England). The resin blocks were sectioned axially to the defect using the ground sectioning technique described by Donath and Breuner19 for hard tissue. The histologic sections were stained with Stevenel's blue and Alizarin red S.
Histologic slides nearly 100 μm thick from each defect were divided into thirds (apical, mid ,and cervical) and the images were captured using LEICA DC 300F video camera (Leica Microsystems GmbH Nussloch, Germany) coupled to a LEICA MZFL III stereomicroscope (Leica Microsystems GmbH Nussloch, Germany), at ×100 magnification. The images were analyzed using Image J software (National Institutes of Health, Bethesda, MD) through the point-counting method described elsewhere.2,4,5,20,21 A grid with 56 intersection points was superimposed on each third, 168 points for the entire defect (Figure 3). The number of points was converted into percentage of the area for bone, biomaterial, and bone marrow.
X-ray diffraction and scanning electron microscopy analysis
These two analyses were aimed at characterizing the tested HAs so as to correlate biomaterial composition and morphology with the respective biological behavior. The x-ray diffraction (XRD) spectra were taken using Cu (kα1) radiation (Siemens D5000 diffractometer, Germany). The spectra were recorded from 2θ = 6 to 60 at a step size of 0.02°. A Zeiss DSM 940A (Germany) microscope was used for the scanning electron microscopy (SEM) analysis.
At 6 and 12 weeks, the 2 types of HAs in 2 different granulation each and the control were compared in relation to the amount of material (HA groups), bone, and bone marrow occurring in the defects. The Friedman test was used for multiple comparisons and Wilcoxon's test was applied when statistical differences were identified. When the data were compared, focusing on the differences between 6 and 12 weeks, the values of material, bone, and bone marrow were compared in pairs using the Mann-Whitney U test. Finally, the Friedman test (for multiple comparisons) and the Wilcoxon test (for paired observations) were applied for data of bone density at the apical third of the defects comparing the HAs and the control groups at 12 weeks. A P value of ≤.05 was considered significant.
All surgical sites healed uneventfully, and there were no signs of clinical reaction to the treatments used for this investigation.
The humerus cross sections at both 6 and 12 weeks showed tightly packed HA particles all over the area of experimental defects. Invariably, the bone tissue was in intimate contact with the particles, drawing a mosaic-like arrangement (Figure 4). Several trabeculae could be seen spanning the cortical-cancellous bones from contiguous areas into the defect. Osteoblast layers were lining the woven bone at both 6 and 12 weeks (Figure 5) suggesting that the process of bone deposition was yet in its course. Osteoclasts in Howship's lacunae at bone surface indicated evidences of bone remodeling in 6 weeks (Figure 5). Part of the surface in some particles was closely associated with areas of connective tissue with variable density of collagen matrix. In these areas, foreign body–like multinuclear giant cells were observed in close contact with the surface of the particles (Figure 6).
In control groups the new bone formation process resembled an attempt to restore the original structure. Nevertheless, the bone formed at the cortical region showed characteristics of cancellous bone, characterized by fine bone trabeculae. Moreover, the bone at the open edge of the defect was filled with dense connective tissue that, in some cases, extended into the mid third creating a concave surface (Figure 7).
Especially at the apical region, the defects filled with HA were mostly occupied with thick trabecular bone, contrasting with a less trabecular pattern in areas adjacent to the defect (Figure 4).
Considering the bone density of the apical third (Table), the comparison among the 5 groups presented significant difference in relation to the bone formation (P = .009). The bone amount of control group was statistically lower in relation to BHA(s), BHA(l), and SHA(s) (P < .05). Besides that, BHA(s) was higher than the 2 synthetic ones and BHA(l) was higher than SHA(l) (P < .05). Considering the levels of bone marrow, the control group presented considerably higher amounts (P = .05) compared with the remaining materials, and SHA(l) was related with larger areas than SHA(s) (P = .05).
Analysis of XRD
Given that the XRD is a granule-size independent method to measure phases of compounds (crystalline or amorphous), the diffractograms were presented for BHA and SHA irrespective of granule size (Figure 11). The diffractograms showed that only the HA phase was present in the samples. The BHAs and the SHAs presented the sharpening of the main reflections located between 2θ = 30–35°. This is compatible to HA with highly crystalline structure, according to the 09–0432 form22 and to Deligianni et al.15
Analysis of SEM
The photomicrographs obtained from HA powders dramatized the differences in relation to the size of the granules, according to the one established in the samples, as well as a similarity in the format. When the granules were evaluated in larger magnification (Figure 12), it was possible to observe a difference in relation to the surface roughness of the granules. In all samples, the granules had micropores (<10 μm),23,24 but the surface area of the granule was clearly larger in the BHAs because of its greater surface roughness.
This study compared BHAs and AHAs with different granules size as bone substitutes for the healing of cortico-cancellous bone defects in the dog humerus. The HAs' composition characterization through XRD analysis as well as the relative standardization of granule size of SHA and BHA used in this study ensured an overall understanding about the role played by both origin and particles size on the osteoconductivity of these bone substitutes. Moreover, this investigation gains further importance because of the scarce scientific documentation in the literature over the comparison between BHA and SHA either in vitro or in vivo.
With regard to bone formation, the outcomes of HA implantation after 6 weeks showed no consistent effect in the comparisons. At this experimental time the HA granules were captured paving the whole extent of the defect with no evidence of inflammatory cells infiltrate. At 12 weeks, the amount of bone tissue found in defects treated with BHA(s) was significantly higher than at 6 weeks. Compared with SHA(s) and SHA(l), BHA(s) sites were related with significantly larger bone areas at 12 weeks. Also, the amount of bone formed in BHA(l)-treated defects was consistently higher than in the SHA(s)-treated defects at this experimental time. In fact, the SHA(s) delivered the least bone amount compared with other groups (control included).
There is some scientific controversy about the size of HA granules that would not come to harm bone-forming cells differentiation and proliferation. In an in vitro study, Evans25 found increased fibroblasts mitotic rate in the presence of synthetic HA particles between 3.7 and 99 μm. Oonishi et al,8 in an in vivo study, demonstrated that particles larger than 10 μm were required for successful bone augmentation procedure. Sun et al9 tested different sizes of synthetic HA particles (0.5 to 3.0, 37 to 63, 177 to 250, and 420 to 841 μm) in osteoblast culture and demonstrated that this cell population grew at a similar rate on the HA 37 to 841 μm particle size range, contrasting with the poor results obtained with 0.5 to 3.0 μm granules. Weissenboeck et al26 examined mesenchymal stem cells in relation to the ability for osteoblastic differentiation and the output of proteins related to bone formation cultured on algae-derivative HA granules (10 to 100, 200 to 500 and 600 to 1000 μm). The authors concluded that the smallest granules (10 to 100 μm) were more effective. In our study, the BHA(s) granule size ranged from 150 to 200 μm and the outcomes in terms of bone formation at 12 weeks (47.22%) were superior (23.10%) to those reported by Artzi et al20 with BHA granules ranging from 250 to 1000 μm in defects made on the mandible of dogs using the same size bone defect as in our study. When it comes to clinical application, BHA is commercially available in granules ranging from 250 to 1000 μm.2 All these data together may indicate a lack of consensus in the literature toward the most suitable HA granule size for bone substitution. The present study showed that BHA(s) performed better than SHA(s) and SHA(l) at 12 weeks, but not against BHA(l). Intriguingly, SHA(s) exhibited the poorest results even when compared with SHA(l) and the control. One can raise the question as to whether granules size within a determined range or the biomaterials themselves were responsible for the observed effects.
Indeed, in our study the implantation of BHA resulted in significantly more bone formation in the defects than SHA implantation, when the data were taken irrespective of the granule size (P = .03). These findings are supported by an ultrastructural comparative study between bovine and synthetic bone substitutes using 2- and 3-dimensional images derived from SEM and transmission electron microscopy.27 The authors remarked that the BHA after deorganification still conserves a template reflecting mineral-collagen interaction and crystallites. This structural characteristic of the bone mineral phase is not demonstrated by SHA materials, which are formed without a guiding organic template. Others suggested, based on an in vivo study,28 that BHA supports the maturation of collagen type I acting like a mineral deposit, which delivers ions for the new osteoid and promotes mineralization.
The surface texture is regarded as an important factor in the osseoconduction by the HAs. The SEM analysis showed that all samples tested in this investigation presented micropores. Andrade et al29 studied dense and porous HA cylinders and observed the fibrous tissue development surrounding dense implants and the direct contact of the new bone formed in the porous implants. Takeshita et al14 used dense HA granules (300 to 600 μm) in bone defects created surrounding osseointegrable implants and reported fibrous encapsulation of the granules. They concluded that dense HA granules negatively interfered with bone formation. Many other studies have reported improved bone-HA integration when the particles presented micro- and macropores.1,2,7,30,31 Recent studies have demonstrated that the increased HA surface roughness positively affected osteoblast adhesion and proliferation.15,31 In our study the BHA exhibited a rougher surface than SHA, which may also have cooperated for the best results in terms of bone formation in sites implanted with BHA.
Although bone formation was evident in all tested groups, the present study failed to demonstrate significant resorption of BHA and SHA granules from 6 to 12 weeks of the experiment. Ideally, the bone substitute should conduct or induce bone formation at the same time it is completely resorbed and substituted by bone tissue.2 Evidence from previous studies suggests that HA resorption can be mediated by cells (degradation by macrophages and osteoclasts) or by disintegration through the action of the extracellular fluids (chemical dissolution).8,28,32 Briefly, in our study the histologic analysis showed the presence of foreign body–like multinuclear giant cells in close contact with the HA surface and bone formation adjacent to the particles. Liljensten et al3 stated that even for HAs considered absorbable, the resorption process is slow and the finalization is not well determined. The same authors, when comparing absorbable and nonabsorbable HAs over 12 weeks, did not observe statistical differences between granule areas either. Artzi et al,1 who evaluated BHA in dog mandibles using the same defect size as in our study, reported that despite the excellent osseoconduction of the material it was present at the experimental sites with no substantial resorption after up to 2 years of remodeling. A great deal of studies have reinforced similar behavior of HAs.2,5,7,31,33,–35 In the context of osseoconduction output, the higher the crystallinity of the HAs the less likely that the granules will break down into fine bits and the better will be the ossointegration.8,9,25 The likely explanation for the poor resorption rate of HAs in our study might be found in the XRD outcomes. Both BHA and SHA were highly crystalline and, therefore, less soluble and more resistant to granular disintegration by extracellular fluids.32,36 This finding must be balanced with the trend that compounds formed with little crystalline and much amorphous (ie, HA/tricalcium phosphate at 30:70 ratio) bioceramics leads to increased bone regeneration and biomaterial replacement by bone tissue than crystalline HA alone.37
The finding that test and control groups failed to present significant differences at 6 and 12 weeks can pose doubts whether or not 5.0-mm diameter defects in the dog humerus were actually of critical size. The sizes of experimental circular bone defects reported in the literature for HA testing are 5.0 mm in rabbit tibia;3 6.4 mm, 6.0 mm, and 4.3 mm in rabbit femur;8,10,23 5.0 mm in dog mandibula;1 3.0 mm in dog femur;32 2.0 mm in rat femur;29 and 8.0 mm in rat calvarium.30 Given that histologic analysis has associated the healing in control defects with a concave configuration, which differs from most of the test groups in which the bone formed in the defects was flush with the neighboring cortical bone, it is likely that wider defects could also lead to significantly reduced amount of bone in the control groups. Nevertheless, in our study the comparison between test and control groups did not reveal any significant difference regarding bone formation until 12 weeks. These outcomes are in agreement with Artzi et al,1 who evaluated porous BHA in the dog mandible using the same size bone defect as ours and observed that a time frame of 12 months (48 weeks) was needed before the test group began to exceed control's bone amount in a statistically significant level.
The bone marrow area decreased in all sites treated with biomaterial compared with control sites. This seems an obvious consequence of space occupation by HA granules in experimental defects. An important observation was, however, that the BHA(l) group showed significantly smaller bone marrow area at 6 weeks than the others. This finding may lie behind the tendency to an increased amount of bone deposition by BHA(l) at the same period, despite the lack of statistic significance.
A noticeable finding in the histologic analysis, also confirmed by the histomorphometric analysis, was the increased bone density found at the apical region of defects filled with the HAs compared with the adjacent area. Although normal cancellous bone of the dog humerus appeared with large bone marrow areas and sparse/fine trabeculae, the presence of BHA, for example, led to statistically higher levels of bone formation at 12 weeks (BHA[s], 50.60 ± 8.40; BHA[l], 37.50 ± 5.04; and control, 28.34 ± 2.94). These results, especially with (BHA[s]), are consistent with those of Artzi et al,2 who found averages of 51.5% to 63.9% of bone tissue and 27.6% to 18% of connective tissue in the apical region of human extraction sockets. This increase in bone density seems relevant to the preservation of postextraction alveolar ridges, especially in areas of bone type III and IV (Lekholm and Zarb's classification38), intended for furthermost titanium implants placement.28
Comparing the histologic features of the control group and test groups, the former was always related with a concave formation at the defect's open end. The particles were intentionally compacted at the insertion in the test groups, which may have led to the formation of a physical barrier that prevented the soft tissue from penetrating into the defect. Similar findings were found in the literature,1,2 showing that the association with membrane for guided bone regeneration produced no statistical difference concerning the volume of regenerated bone as well as the external bony contour compared with the use of biomaterial alone.20,39 As to the suitability of BHA and SHA grafted sites as a bed for titanium implants, previous studies reported that 20% to 30% of the implant surface keeps in tight contact with the particles of the biomaterial.2,20,40 Despite the fact that BHA and SHA dramatically reduced the bone marrow area in our study and, consequently, increased the mineral content in the defect, the use of bone substitutes can potentially jeopardize the osseointegration of titanium implants. Considering the high crystallinity of these biomaterials and, consequently, their diminished resorption rate, one can expect complex titanium-HA interface. This sort of interaction tends to result in biomechanical instability given the combination between resilient (bone) and rigid (biomaterial and titanium surface) materials. Well-controlled experimental studies are required to test this hybrid interface under implants in load-bearing condition.
To our knowledge, this is the first in vivo study to address the comparison between BHA and SHA materials focusing on the influence of granule size on osseoconduction. The outcomes indicate that BHA performed better than SHA as osteoconductive biomaterial, as assessed by the amount of bone measured in the implantation sites at 12 weeks. The BHA's distinctive characteristics, such scaffold structure and surface roughness, rather than the granule size itself, accounted for this improved biological behavior. Neither BHA nor SHA exhibited signs of resorption until 12 weeks of implantation, which may relate to the high crystallinity presented by these biomaterials.
We would like to thank Eliana Cristina da Silva Rigo for her assistance with the scanning electron microscopy and x-ray diffraction analysis and Sebastiao Carlos Bianco for histology slides preparation. This study was partly funded by the Turkish Republic Government Planning Organization in the framework of the project, “Manufacturing and Characterization of Electro-Conductive Bioceramics,” #2003 K120810.
Alexandre L. Carvalho, Paulo E. P. Faria, Marcio F. M Grisi, Sergio L. S. Souza, Mario Taba, Jr, Daniela B. Palioto, Arthur B. Novaes, Jr, and Luiz A. Salata are with the Department of Oral and Maxillofacial Surgery and Periodontics, Faculty of Dentistry of Ribeirão Preto, The University of São Paulo at Ribeirão Preto, SP, Brazil. Address correspondence to Dr Luiz Antônio Salata, Department of Oral and Maxillofacial Surgery and Periodontics, Faculty of Dentistry of Ribeirão Preto, The University of São Paulo at Ribeirão Preto, SP, Brazil. Avenida do Café, s/n, 14040–904, Ribeirão Preto, SP, Brazil (e-mail: email@example.com).
Alexandre F. Fraga is with the Department of Materials Engineering, Faculty of Chemistry, The Federal University of São Carlos, São Carlos, SP, Brazil.
L. Sevgi Ozyegin is with the Department of Dental Technology, Vocational School of Health Related Professions, and Faik N. Oktar is with the Department of Industrial Engineering, Faculty of Engineering, Marmara University, Istanbul, Turkey.