The influence of the chosen in vitro bone simulation model on intraosseous temperatures and drilling times

There is no consensus about the most suitable in vitro simulating material investigating heat generation during bone preparation. The aim was to compare heat increases and drilling times of bone removals in different bone simulating materials and to compare them to fresh human cadaver bone. A cavity was drilled in the following samples: (1) bovine rib; (2) pig rib; (3) 20 PCF (lb/ft3) polyurethane (PU) block with 3 mm (50 PCF) cortical layer; (4) 20PCF PU without cortical; (5) 30 PCF PU with 2 mm (40 PCF) cortical; (6) 30 PCF PU with 1 mm (40 PCF) cortical; (7) 30PCF PU without cortical; (8) poly-methyl-methacrylate (PMMA); (9) fresh human cadaver rib. Data were analyzed with ANOVA followed by Tukey’s post hoc tests. P < 0.05 was considered significant. Highest heat increases and slowest drilling times were found in bovine ribs (p < 0.001). Regarding temperatures, human ribs were comparable to the pig rib and to PUs having cortical layers. Considering drilling times, the human rib was only comparable to the 20 PCF PU with 3 mm cortical and to 30 PCF PU without cortical. By the tested in vitro bone removals, only the 20 PCF PU with 3 mm cortical was able to simulate human ribs, considering both temperature increases and drilling times.

but they detect only a single temperature point and need preparing pilot holes. Infrared thermography detects an overall thermal profile including also the measurement of leaking heat; however, it detects only the surface temperature, without predicting the temperature of the actual drilling site 1,2,6,11 . According to Möhlhenrich et al. 's systematic review, 18 studies used thermocouples, and 7 used thermal cameras of the involved, analyzed studies 12 .
Since, there is no universally recommended in vitro model, and researches use plenty different models, the comparison of different study results are almost impossible 12,13 . Moreover, there is very limited information, on how the choice of bone model influences outcomes by similar bone removals 13 . As Fernandes et al. (2017) wrote, to date just a relatively few numbers of papers have directly compared heat development of drilling in biological tissues with synthetic bones 1 . According to our best knowledge, this is the first attempt to compare porcine and bovine ex vivo models simultaneously with different density polyurethane blocks and with fresh human cadaver bone by the same bone drilling procedure (common in oral surgery bone removals 3 ).
The purpose of this in vitro study was, to investigate a standardized bone drilling's thermal effect and the duration of these preparations in different, frequently used bone simulating materials and in human bone to estimate the effect of the chosen model on the results.

Materials and Methods
Experimental set-up. The tested bone removal in our study was an average and everyday bone preparation method in oral surgery 14,15 . Tungsten carbide round drills (HM141A, Hager & Meisinger GmbH, Neuss, Germany) with a diameter of 3.1 mm were used at 8000 revolutions per minute (rpm) speed, with 6 N of axial loading and 60 mL/min physiologic salt irrigation volume. A special testing device was able to standardize drilling parameters (speed, axial load), and a perpendicular, 5 mm in depth cavity drilling 14 . The device cooperated with a physiodispenser unit (Implantmed SI-915, W&H, Bürmoos, Austria) and a surgical straight handpiece (SL-11, W&H) [ Fig. 1].
In every of the below detailed groups, 24 drillings were performed, i.e. altogether 24 × 9 = 216 test cavity preparations. All drillings were performed by the same operator (first author), who had remarkable experiences with the drilling tower illustrated in Fig. 1. After twelve drillings, the drill was disposed and a new drill was applied, to disclose the unwanted effect of wearing.
Set-up for temperature measurements. Experiments were performed in an air conditioned room, at standardized room temperature of 24 °C. Temperature measurements and heat registrations were performed by A metal template determined the places of the thermocouple probes in a standard, 1 mm distance from the test cavities, simultaneously allowing placements into the same depth of 5 mm, under the bone surface (Fig. 2). The cavity of the probes was drilled with a 0.5 mm in diameter fissure drill (203 RF, Hager & Meisinger GmbH) ( Fig. 2a,b). Probes were isolated with rubber tubes [prepared from 22-G wing needles, B. Braun Melsungen AG, Melsungen, Germany], to disclose the disturbing effect of the cooling liquid (Fig. 2c).
Statistical analysis. The statistical analyses were performed with SPSS v. 22.0 (SPSS, Chicago, IL). The Kolmogorov-Smirnov test was applied to test the normality of the distribution of the data. The changes in temperatures and drilling times were compared between the tested experimental bone model groups with one-way ANOVA, followed by the Tukey's HSD post hoc test. P values below 0.05 were considered significant.
Ethical approval. The Regional Ethics Committee of University of Pécs approved the study (7621/ PTE/2018). Animals were not sacrificed for the experiment; bone samples were bought from the butcher.

Results
Mean temperature increases of the drillings (see Fig. 3. and Table 1) were significantly the highest in bovine ribs (p < 0.001). No other investigated model was able to reproduce these temperatures. Temperature increases in human ribs were comparable entirely with polyurethane blocks with cortical layers, or the pig rib (Table 2). 20 PCF PU (p < 0.001) and 30 PCF PU (p < 0.001) showed significantly less heat productions, than human bone. In addition, heat increases in the cancellous PU models were significantly lower, than in PU blocks with cortical layers. Regarding average temperature increases, the 20 PCF and 30 PCF cancellous blocks functioned similarly (p = 0.968).
Preparation times showed (see Fig. 4 and Table 2), that the duration of drilling in bovine rib was significantly the longest (p < 0.001). The pig rib was comparable to Nacional Ossos PU blocks, laminated by 2 or 1 mm cortical layers (p = 0.335 and p = 0.973). Interestingly, the Sawbone PU block with the 3 mm cortical layer was prepared significantly faster, than the pig rib [p < 0.001]. Drilling times in the human rib were comparable to the Sawbone PU block with the 3 mm cortical layer and to the cancellous 30 PCF polyurethane block.
Comparisons of the human bone to the animal bones and PU blocks are represented in Table 2, while eight randomly chosen graphs of drilling temperatures of the study groups are represented in Fig. 5.

Discussion
Probably the use of drills or other rotating instruments is still the most frequent way of bone removal in the field of oral surgery. Surgeons' aims are to reduce operation time and patients' discomfort and to keep thermal parameters acceptable, however an excessive irrigation may disturb the optimal visualization 14 . In contrast, inadequate irrigation or suboptimal drilling conditions may disturb bone regeneration processes and compromise wound www.nature.com/scientificreports www.nature.com/scientificreports/ healing 6 . The surgeon should apply optimal drilling parameters (revolutions, axial feed, method of irrigation etc.) and chose the correct drill specification (drill diameter and design, drill material, drill wear etc.) to regulate intraoperative temperatures 4,11 .
Intraosseous heat depends on the quality of bone, such as bone density or thickness of the cortical layer 6,16 . Unfortunately, the use of fresh human cadaver bone is limited in the literature, while an optimal, or universally applied or accepted bone substitution model does not exist. Several investigations use an in vitro design, examining different animal or synthetic materials to simulate human bone. Animal bones originate usually from pig or bovine -including ribs, mandible, femur, iliac crest-, rarely from sheep. Artificial materials include polyurethane (PU), poly-methyl-methacrylate (PMMA) or hydroxyl-apatite and in one study resin and hardwood was used. Basically, the in vitro design has several drawbacks and limitations. There is no blood circulation in the bone or in the model, body temperature can be set and held with difficulties and the parallelism with human in vivo characteristics depends highly from the chosen in vitro model. While animal bone samples originate from living creatures, which should be a benefit, features change from one individual to the other, exhibiting a variable structure. Since ex vivo bone specimens consist of organic and inorganic materials forming a complex anisotropic tissue, it is difficult to investigate the thermal properties. To avoid such variabilities, synthetic polyurethane foam blocks with different densities were introduced 1,2 . This engineering material provides a uniform testing model, showing consistent properties and less susceptible to failures, which require less repetitions 1,2 . Furthermore, the handling and storage of synthetic materials may be much easier.
In case of animal bone models, regarding composition (ash, hydroxyproline, and extractable proteins, IGF-I), density (BMD) or mechanical properties (e.g. fracture stress) different species seems more optimal to simulate human bones than others 17 . Considering before mentioned parameters together, the dog's bone characteristics seems optimal, while rat's parameters seems the least, according to Aerssen et al. 17 . In in vitro investigations, the usage of bovine and porcine or synthetic bones are dominant.   17 . Supposedly, this difference in BMD values might be represented in the significant differences between bovine and pig ribs or between bovine and human ribs in the current study. Regarding thermal characteristics, considering thermal conductivity and specific heat capacity, interesting differences and similarities are seen between human, animal and synthetic bones (Table 3). Data show several times an overlap with humans, but the correct specimen selection might be frequently random and accidental. According to Zdero et al. (2016) the 60% of heat is dissipated into bone chips during drilling, while 40% is absorbed by the host bone, so when e.g. a PU synthetic material forms different chips (PU forms rather a kind of powder in our opinion, see Fig. 2c) because of the missing organic compounds, it can hardly mimic the human bone perfectly 19 . When a synthetic material do not form such adhering 'biological material' characteristic bone chips, the possibility of chip-clogging (which is a significant temperature increasing factor) of the flutes is reduced 20 . That may partially answer, that temperature increases are not only influenced by similar densities or thermal parameters. Additionally, as Davidson and James concluded, heat capacity (i.e. specific heat) is more important parameter of the material as heat conductivity 21 . While changing the thermal conductivity had a negligible effect on intraosseous temperature, a 10% increase in heat capacity resulted in a 5% drop in temperatures 21 .  www.nature.com/scientificreports www.nature.com/scientificreports/ Another interesting difference was found in mammalians' bone histological structure, i.e. the interspecies differences of Haversian canal diameters, however, its influence on the heat dissipation is obscure. While human Haversian canals' diameter were found to ~100 µm, that value was in pig and in bovine only ~35 µm 22 .

Comparison of investigated drills
According to manufacturers' intentions, the wide variety of densities -i.e. the PCF scale usually from 10 to 50 of PU blocks-should cover the broad spectrum of human bone characteristics. The selected polyurethane blocks in the current study, i.e. from 20 to 50 PCF, with (1-3 mm) or without cortical layers, were selected intentionally, to investigate the simulating potentials of the PU material. The 20 and 30 PCF PU with corticals of 40 and 50 PCF are offered in the literature usually to simulate the typical quality oral bone (D1 and D2 densities according to the classification of Misch) 5,6 . According to the literature, the density of the PU reflects to the human bone densities as follows: D1 = 0.48-0.64 g/cm 3 = 30-40 PCF; D2 = 0.32 g/cm 3 = 20 PCF 5,6 . According to our results, PU blocks with cortical layers showed similarities with human and pig ribs, but bovine's rib showed significantly different results (Figs 3-5). Also, an interesting difference was seen in the drilling times between the two manufacturers who produced PU cortical blocks (Sawbone vs Nacional Ossos). Although Sawbone blocks had thicker cortical (3 mm vs 1 mm and 2 mm) and a higher density of the cortical (50 PCF vs 40 PCF) they needed significantly shorter drilling times. Further model studies need to be conducted to explain that phenomena. Fernandes et al. (2016) investigated four PUs with different densities 2 . Cortical was simulated by 50 PCF and 5 PCF blocks and cancellous bone by 20 PCF and 7.5 PCF blocks. It was clear, that temperatures were strongly correlated with densities, but direct comparison with that study is difficult because of the very different drill specifications and drilling parameters, moreover only the drilling in 50 PCF material was cooled with air stream, which is unusual in oral surgery. While mathematical (describes usually an overall cutting force) and numerical (simulates both forces and material responses) modeling is straightforward, and finite element (FE) analysis can be considered as a solution to estimate the parameters of drilling in order to prevent the problem of osteonecrosis 23,24 , it was interesting to see, that between numerical and experimental temperature increases there was a difference of up to ±2.0 °C, depending on the density of the PU blocks. In a study of Tu et al. (2013) the peak temperature difference from the proposed FE model and the experiment was no more than 3 °C 25 . As Mediouni et al. (2019), recently stated, "software tools do not replace experimental testing", but "they provide a valuable and rapidly evolving option" 24 .
Unfortunately, direct comparisons of bone removals in different studies is very difficult, because investigations usually use different bone models, drilling parameters, drill designs or materials. Furthermore, studies using similar bone removals in different bone simulation materials is infrequent 4 . For instance, in two investigations, very similar implant drills were tested (stainless steel 4.2 mm and 4.3 mm in diameter) with similar drilling parameters (20 N axial loading, 1500 rpm) 8,26 and with similar method of temperature measurements (thermocouple probes). An important difference was, however, the bone simulation material, i.e. bovine femur vs. synthetic polyurethane blocks. The first found an average of 32.1 °C, while the second only an average of 22.2 °C of temperature elevations during implant bed drilling. That was a 30% difference in results. Our results can confirm such a high difference (bovine: 3.85 °C vs PU: 1.63 °C, i.e. ~40% difference), although earlier mentioned researches investigated implant bed drilling with ~12-13 mm deep cavities, while our test cavities were only 5 mm deep. Interestingly, a recent investigation showed, that temperatures during drilling were significantly higher in bovine femurs, than in human cadaveric tibiae 27 . The proportion of temperature increases in human and bovine samples were around 1: 2.65 in the previously mentioned study, while in ours this was very similar, 1: 2.53.
In contrast, in two other investigations both researches investigated similar piezoelectric preparations, but in the first it was performed in bovine ribs, in the second in pig mandibles, and they found very similar average temperature elevations (~3.3 °C vs. ~3.5 °C) 15,28 . It was clear during our experiments, that not only the storage, but the customization of PU blocks into the bone fixation box was much simpler than in case of animal or human bones. Storage of fresh bones is limited in time without freezing, while frozen bone after unfreeze might change its' structure and physical properties. Moreover, in case of human bone, several ethical and infection control issues should be managed.
A frequently mentioned drawback of in vitro bone simulations is the missing blood circulation, as circulating blood may help to dissipate heat. According to several authors, however, it has less significance on temperature accumulations 11,29,30 . Another noted drawback is the missing physiologic body temperature. Several attempts are seen in researches to simulate ~37 °C body temperature. In some of them, the bone is heated to 37 °C before experiments, in others the room is heated and kept on 37 °C or experiments use a sinking bath [container with preheated physiologic salt], with or without a thermostat. Some opinions, however, suggest that bone sample in a sinking bath may dissipate much less heat, than it would happen in vivo, furthermore thermocouple probes may be disturbed by the fluid of the bath 4 . According to Lee et al., the insistence to the real 37 °C, as the initial temperature of experiments, can not be proven 31,32 . They showed evidence in two of their researches, that intraosseal temperature changes (ΔT) were similar, when initial temperatures were 26 °C or 37 °C.
Although, PMMA occurs in more investigations as bone simulation model, according to our experiences it can not be compared with the pig or bovine ribs or the PU. When the places of the thermocouple probes were drilled, we experienced much higher resistance of the material. During drilling, at 8000 rpm and 6 N axial loading, the drills were unable to penetrate into the material. After then, the axial pressure was elevated. At approximately 38-40 N pressure, the drills were able to drill the PMMA, but they needed more then 23-26 seconds to reach the 5 mm depth. During this period, PMMA melted and surrounded, coated the drills (Fig. 6). The removal of the acrylate from the drill was very time consuming. Probes measured an average of 18 °C heat increase during drilling.
Considering the very different drilling and heat increasing conditions of PMMA, authors did not involve these data in the statistical analysis. In contrast, according to Pandey and Panda's orthopedic simulations 33 , PMMA was prepared with 6-8-10 mm in diameter spiral drills at 2500 rpm, with a feeding rate of 35-45 mm/min successfully, so it can be supposed, that PMMA preparation is possible, however, not for every bone removal simulations. In fact, it is a frequently experienced difference, that orthopedic and trauma related drillings use significantly higher axial pressures than oral and maxillofacial surgery related maneuvers (80-200 N vs. 3-25 N) 3 .

Figure 6.
Drilling the poly-methyl-methacrylate (PMMA) with the investigated bone removal parameters was not successful. When axial loading was increased, melting of the PMMA occurred, which ruined the drill. In the right upper corner, the applied drills can be seen after drilling the polyurethane versus the PMMA.