A customized 3D-printed histological microgrinder for the study of metallic endoprostheses

ABSTRACT

Use of histology is the key when evaluation of bone and soft tissue integration of any implanted metallic prosthesis is required. This relies on the ability to prepare very thin sections by grinding down resin embedded samples. Manual grinding has historically been used with variable success, and thus, a number of commercial microgrinders have been previously marketed, however, at a significant cost. The following describes a practical method to 3D print and build a microgrinder construct retrofitted to a metallurgic wheel grinder/polisher previously available. The design files are also supplied, which allow one to implement customized modifications for virtually all types of wheel grinders/polishers circumventing the need to procure highly costly appliances. Recommendations are included on how to safely and reproducibly prepare microscopic sections using the described construct.

KEYWORDS:

• Endoprosthetic
• ground section
• histomorphometry
• implant
• 3D printing

Introduction

The advent of additive manufacturing technology in the recent years has not only revolutionized goods manufacturing methods but also made it possible to custom-design and build large patient-specific metal endoprostheses, where radical resection of diseased bone is required, i.e. endoprosthetic reconstruction. New legislative restrictions on such devices have been implemented in many countries prompting the need for urgent research into their effectiveness, safety, and further improvements.

Large metal prostheses pose unique challenges for researchers involved in their evaluations. Computed/microcomputed tomography of such composite samples is almost always associated with photon starvation and beam hardening artefacts sometimes rendering the resulting data completely useless. Furthermore, x-ray-based methods are of limited value when evaluation of soft tissue integration, an emerging area to be addressed particularly with larger maxillofacial endoprostheses, is needed. This highlights the significant role of histological techniques in the scientific evaluation of these applications.

Historically, manual grinding and polishing of bone and metal samples has been the main method for obtaining thin enough sections from such samples for the purpose of light or fluorescent microscopy. This involves manually holding a section attached to a slide against a moving abrasive surface, which is usually a rotating wheel. The section should be kept parallel to the abrasive surface at all times, and as the sample becomes thinner, the consequence of any mistakes will be more detrimental. With larger samples, this technique is highly challenging if not completely impossible. Various appliances have been designed and marketed to address this issue and to improve accuracy, nonetheless, these appliances are highly costly and thus not widely available to the broader scientific community [1–3].

This manuscript describes and provides the design and method of use of a 3D printed microscopic slide holder and associated construct to achieve planar grinding of metal and/or undemineralized bone samples. The 3D files are also provided which may be modified to suit previously available metallurgic grinders (as described here) or grinders that can be easily purchased from many retailers.

Materials and methods

Wheel grinder

The construct described here was designed to retrofit a previously available 600W 2011 model FORCIPOL 1 V wheel grinder-polisher (Metkon Instruments Inc. Bursa, Turkey). The unit has a 250 mm rotating wheel within a round housing measuring 308 mm in diameter (Figure 1a). The boundary of the housing has a ledge that normally accommodates a plastic rim to prevent splashing of water and debris (splash guard). This rim needs to be removed to allow for the attachment of the microgrinder construct described herein (Figure 1b). The design described can be easily modified to fit virtually any wheel grinder/polisher. Relatively inexpensive grinders can be found and procured from a number of online retail stores.

Figure 1. (a) The planar microgrinder was designed based on a FORCIPOL 1 V wheel grinder/polisher. The plastic splash guard has been removed exposing a ledged boundary to accommodate the microgrinder construct. (b) The FORCIPOL 1 V wheel grinder fitted with the planar microgrinder construct during operation.

Block trimming and initial sectioning

An IsoMet low-speed saw (Buehler, Illinois Tool Works Inc, Illinois, USA) fitted with an IsoMet precision blade (IsoMet Blade, 15HC, 127 mm) was used to trim the blocks and cut the sections to be ground and polished. Most sections were cut at 700 RPM with little to no weight placed on the swivel arm, and tap water was used as coolant. This set-up may also be used to downsize large bone-metal samples following fixation. A 1:1 ratio of water and ethanol (50%) could be used as coolant for this purpose (70% ethanol is more flammable).

3D design and printing

The parts to be 3D printed were designed in Blender 2.9 [4]. Individual parts were then exported as .stl files using the standard Blender 3D-Print plugin. Parts were then imported into MakerBot Print application and 3D printed in polylactic acid (PLA) at 200 µm layer height using a MakerBot Replicator Z18 3D printer (MakerBot Industries, New York, USA). Some parts had to be re-oriented, and settings re-adjusted by trial and error to obtain the best result.

Mechanical parts and electronics

The 1:19.5 geared electric motor (24 V, 0.6 Amp, 300 RPM) and its mounting bracket were from Greartisan (China). The micro switches (16A/250VAC 3 Terminals Miniature Micro Switch) were from OMRON electronics (Kyoto, Japan). The linear ball bearings (LMF10LUU and LMF12LUU) were from Aexit (China). The SFU1605 ball screw kit (screw shaft, ball nut, motor coupler and the BK/BF12 bearings) were from Sentech (China). The DPDT power relay and socket base kit (DC 24 V Coil Power Relay 12A DPDT LY2NJ 8 Pin plug in) and titanium rods (10 and 12 mm in diameter) were purchased from eBay (California, USA). All screws, bolts, and nuts were purchased from Bolt & Nut Australia (Queensland, Australia).

Consumables

Two-millimetre plexiglass (acrylic) sheets can be obtained from many online retail service providers and custom-cut at required dimensions (using a tungsten carbide pen). I recommend using the same length as a standard microscope slide (75 mm). The width, however, should be greater than a standard slide, i.e. 30–35 mm, to allow for more secure attachment to the slide interface and to accommodate larger samples. These can be manually trimmed to the conventional 25 mm width on the wheel grinder prior to microscopy so that slides can fit a standard microscope stage. Glass slides are not suitable as they easily break during grinding/polishing, which can damage the section or cause injury to the user.

Any high-quality waterproof silicon carbide sandpaper (various grits) can be used in the protocol described here. Diamond suspensions and polishing cloths were from Kemet Australia (New South Wales, Australia).

Regular superglue is the most appropriate option for adhering sections to plexiglass slides. I currently almost exclusively use UHU ultra-fast CONTROL (UHU GmbH & Co. KG, Bühl, Germany). Five to six drops (for a 25 × 25 mm sample) placed in the centre followed by slowly lowering a slide over a section or block face generally produces high-quality and relatively bubble-free adhesion. Light-activated adhesives provide the advantage of being able to delay the polymerization until one is satisfied with the mount; however, some of these preparations did not achieve a bubble-free adhesion in the author’s hands likely due to the relatively high viscosity of the adhesives (Figure 2a). Additionally, using such adhesives frequently led to the detachment of sections from slides during later stages of grinding or polishing. Crucially, superglue does not dissolve readily in organic solvents (except for acetone according to the manufacturer) allowing partial deplasticization of the final methyl methacrylate embedded sections in toluene and normal ethanol dehydration (10 min per step). This enables high-quality staining as shown in Figures 2b and 3. The use of light-activated adhesives frequently resulted in loss of sample or wrinkling during deplasticization/dehydration.

Figure 2. (a) A 200 µm thick section of a titanium miniplate fixed to a cadaveric sheep mandible using two 2 mm × 8 mm titanium screws. A section was glued to a microscopic slide using a highly viscous industrial grade UV-activated acrylic glue (Viscosity @ 25 ℃ @ 20 RPM: 30,000 to 55,000 mPa.S) resulting in the entrapment of several air bubbles in between the section and the slide. Miniplate was designed and kindly supplied by MAXONIQ. Scale bar = 5 mm. (b) Stitched photomicrograph of a transverse section through the distal metaphysis of a sheep femur. Section has been obtained at the level of the flange of a titanium prosthesis implanted 12 weeks earlier. Masson–Goldner Trichrome staining. Courtesy of IMCRC Just-in-time patient-specific bone tumour project. Scale bar = 5000 µm.

Figure 3. (a) Fluorescence photomicrograph of a section of a sheep femur containing a titanium screw implanted intra-cortically 12 weeks earlier. Intravital fluorochrome labelling was performed using calcein (yellow to green labels) and alizarin (red labels) administered 2 weeks apart. Darker areas (other than the metal) correspond mostly to either soft tissues or osseous tissues that existed before the administration of the fluorochromes. Scale bar = 500 µm. (b) The same section after staining using the von Kossa and toluidine blue method. Scale bar = 500 µm. (c) Fluorescence photomicrograph of the interfacial region at a higher magnification. Scale bar = 150 µm. (d) The same area after von Kossa and toluidine blue staining. Osteoid (black arrow) can be readily identified interspersed between the darkly stained mineralized matrix and the highly basophilic osteoblasts (white arrow). Here, the cross section of a resorption cavity at the level of its cutting cone can also be seen exhibiting active osteoclasts (yellow arrow) resorbing pre-existing bone (P). R denotes a cross section of another resorption cavity at a level where osteoblastic activity has just started (at its right margin) and the first layer of osteonal lamellae is being laid down. Scale bar = 150 µm.

The epoxy resin used was from AA Composites (Queensland, Australia) and had to be prepared by mixing equal volumes of parts A and B. Any good-quality epoxy resin that has a working time of at least 20 minutes can be used.

Design description

All design files are permanently available on Mendeley (see supplementary material). Parts are labelled in Figure 4a. The slide holder (part A) can freely move in the Z-axis perpendicular to the rotating wheel on the grinder, thus feeding a slide attached to its bottom surface (slide interface) onto the abrasive surface through gravity. The slide interface was manually prepared by filling the preplaced cavity in the 3D printed part with epoxy resin. This was done to take advantage of epoxy’s favourable polishing characteristics (when compared to PLA). This surface was ground and polished once all parts were assembled together and the slide holder was in its final position. This allowed the surface to be ground and polished parallel to the grinding wheel (see under calibration). When wet, the surface tension was usually enough to adhere a slide onto this interface against gravity; however, this was often not sufficient to hold a slide against the rotating wheel during grinding. To ensure reliable slide attachment, the slide interface was designed to have three perforations to communicate with an internal duct system, which connected to a vacuum pump at the top of the slide holder. When turned on, this created necessary force to retain a slide against the rotating wheel.

Figure 4. (a) The schematic of the sample holder construct with various parts labelled (A–G). Red line represents X-axis. Yellow line represents Y axis. BS, ball screw mechanism; LB, linear ball bearings (LMF10LUU and LMF12LUU); LT, L and T shaped handles; MO, electric motor. (b) The position of one of the 3-terminal micro-switches underneath part G. The switch activates when the vertical arm of part B comes into contact with it immediately leading to a reversal of the direction of planar sliding.

Given the inherent porosity of 3D printed PLA, the internal chamber and plumbing needed to be sealed using epoxy resin or silicone. This was best achieved by 3D printing the part with its longer axis vertically oriented and with an infill density of less than 5%. The MakerBot Print application setting was manipulated to reduce the roof surface thickness so that the top surface could be easily removed to access the interior for sealing. This process was delicate and needed to be done in several steps while using a prong or wire to keep the interior plumbing patent while waiting for the resin to set. Another duct system was also incorporated into the slide holder and could be connected to a water source to allow simultaneous cooling of the sample during grinding as well as lubrication (see supplemented video). Some wheel grinders/polishers may be equipped with a water-cooling system rendering this feature redundant.

The mechanism that allowed the central slide holder to move freely up and down (vertical sliding) consisted of two vertically positioned linear ball bearings (LMF10LUU) installed into the two ends of the slide holder. Each linear ball bearing accommodated a 10-mm titanium rod held parallel to each other by parts B and C (Figure 4a). This mechanism allowed the sample to be fed onto the abrasive surface (rotating wheel) while being held parallel to it.

The assembled parts A, B, and C could oscillate in the X-axis (planar sliding) through two pairs of horizontally positioned linear ball bearings (LMF12LUU) in part B. Each linear ball bearing was slided over a 12-mm titanium rod that ran from one side of the base rim to the other (parts D and E to parts F and G). This allowed the slide holder to oscillate back and forth parallel to the rotating wheel. This movement was mainly necessary during polishing (see later under protocol). This requires a ball screw mechanism that automatically reverses the direction of movement upon reaching set points. Part B was designed to have a vertical arm that fits the flange of the ball screw nut, which was secured in place using 6 pairs of screws and nuts (Figures 1b and 4a).

Parts D–F make up the rim and base of the construct, which was designed to fit a FORCIPOL 1 V wheel grinder/polisher in place of the splash guard. These can be modified to fit alternative makes. If it is possible to obtain access to a larger 3D printer than the one used here (i.e. Replicator Z18), the rim can be manufactured as a single part. The base rim also accommodated a ball screw mechanism and the associated electronics necessary for the planer movement of the slide holder. All parts were filled with epoxy resin (where possible) to help enhance durability, particularly where screws were to be used.

To enable easy lifting, a number of L- and T-shaped handles were also included in the design. Alternatively, these can be replaced by commercially available knobs or handles if preferred. If so, the pertinent slots in parts D, E, F, and G can be filled with epoxy to allow insertion of screws. While it should not be necessary to secure the construct onto the wheel grinder due to the construct’s weight and stable design, handles can be used to tie rubber bands or similar for such purpose. The author does not recommend such use as the tension applied on the base rim can deform the construct and interfere with planar grinding despite successful calibration.

For a step-by-step guide on how to assemble the parts, please see assembly instructions provided in supplementary material. Please also note that the choice of PLA for this project was mainly driven by constraints imposed by the 3D printing service used. Therefore, users are encouraged to experiment with alternative materials such as acrylonitrile butadiene styrene (ABS) or various resins if available.

Electronics

A 24-V DC 2.9A power supply was used to power the motor. The power relay circuit functions to reverse polarity, and thus the direction of motor rotation as soon as the slide holder reaches set positions along the 12-mm rod railing enabling back-and-forth oscillation. These positions are determined by strategically placing the two 3-terminal micro-switches underneath parts D and G (Figure 4b). I recommend keeping the distance between the two switches fairly short at about 4 cm to improve planar grinding and avoid the centre of the abrasive wheel (see supplemented video). The circuit (Figure 5) was adopted from CheesyCam DIY Videos and Photography Projects [5]. While this set-up works well, an improvement would be to use an electric motor that has a lower RPM or a higher gear ratio in a bid to minimize the chance of over-polishing even further. Swapping the positive and negative wires on the motor can sometimes help resolve any issues encountered when configuring the circuit.

Figure 5. A circuit diagram demonstrating the configuration of electric components of the planar oscillation mechanism. Adapted from CheesyCam DIY videos and photography projects.

Recommended protocol

Parts of the protocol described here are based on a protocol previously described by Donath and Rohrer (2003) as well as personal communications with Kemet Australia representative [1]. Using the recommended protocol and water cooling at 400 mL/min, working temperatures never exceeded 26.3°C when measured with an infrared thermometer pointed at the sandpaper as it emerged from under the section during operation.

Safety

It is important to follow all local personal safety guidelines when following the protocol described here. Slides can occasionally detach from the slide interface and launch by the rotating wheel. Therefore, eye protection is critical. Hearing protection is also crucial due to the level of noise generated and the lengthy duration of preparation. A reasonably large vacuum trap is needed to prevent water from being sucked into the vacuum pump and to reduce the risk of electric shock. This needs to be emptied at regular intervals.

Section preparation and thickness measurement

For information on histological processing of medical implants, refer to Rousselle et al. [6]. For information on methyl methacrylate embedding of large histological samples, users are referred to Emmanual et al. [7]. The following has been devised based on large methyl methacrylate tissue blocks (~100cc). Original blocks are marked parallel to each other at pre-determined intervals (3-5 mm) and sectioned using an IsoMet Low Speed Saw. The IsoMet cut sections are superglued to a new plexiglass slide and briefly ground and polished (5 min on each p240, p600, p1200, p2500) in static mode (see under planar grinding) to achieve a truly planar face. Each section prepared this way serves as a new tissue block and can be used to generate multiple final sections as required. The thickness of each tissue block (together with the plexiglass slide to which it is attached at this stage) should be measured using a digital micrometre screw gauge and recorded. This needs to be done on two spots spanning the length of the section.

The thickness of a new plexiglass slide is measured at two spots mirroring those on the tissue block and marked on the opposite side of the slide using a tungsten carbide pen. The polished surface of the tissue block is then superglued to this new slide and placed under a 50 g weight for 30 min. Next, the combined thickness of the new slide and the tissue block is measured at the previous two spots, and the glue layer thickness is determined (i.e. subtract tissue block and slide thicknesses from total thickness for each spot).

The tissue block is glued to an IsoMet wafer chuck before being finally released using an IsoMet Saw leaving an approximately 500 µm thick section on the second slide (thinner with smaller samples). The face of the block needs to be briefly ground and polished at this stage if more sections are to be acquired.

As grinding/polishing progresses, sections should be frequently monitored and measured on the exact spots marked on the slide for section thickness to be calculated (section thickness = total thickness – glue layer thickness – slide thickness). Using this method, a final section thickness of 40–50 µm could be achieved in my hands when working with large 25 mm × 25 mm samples. Sections as thin as 20 µm could be obtained with smaller samples (e.g. 5 × 5 mm). Trying to achieve anything thinner was frequently associated with loss of bits from sections especially with metal-bone composite samples.

If backscatter electron microscopy is desired, section thickness will no longer be relevant. It is sufficient to simply superglue a section with its surface of interest on top onto a plexiglass slide.

Calibration

An initial calibration step is required before the first use or when planar grinding cannot be achieved any more (every few months in author’s hands with heavy use). To do this, the electronic oscillating mechanism should be turned off (static mode), and the slide holder position should be manually adjusted by rotating the ball screw to avoid the centre of the abrading wheel. It is paramount to always use this same position during all later static grinding sessions. This step involves grinding and polishing the slide interface such that the final finished surface is parallel to the grinder wheel and is smooth enough to securely hold a slide against the rotating abrasive wheel when vacuum pump is in operation. Start on a p80 silicon carbide sandpaper until the surface is fully planar (parallel to the wheel). Several sandpaper changes may be required to achieve this. Once planar, sequentially move on to p240, p600, and finally p1200. Spend less and less time grinding as the sandpaper becomes finer in order to avoid rounding of edges (over-polishing).

Planar grinding

This is the first step in reducing the thickness of a section and must be performed in static mode. Planar grinding is best achieved on p80 silicon carbide sandpapers followed by p240 and p600. Most of the grinding should be performed on coarser sandpapers rather than finer ones to save time and prevent over-polishing (non-planar sections or round edges). Grind on p80 at 400–500 RPM until a thickness of 250–300 µm is achieved (change sandpapers at least every hour). Grind on p240 at 300–400 RPM until a thickness of 180–200 µm is achieved (one sandpaper is usually enough for 4–5 slides). Grind on p600 at 200–250 RPM until thickness is just below 130 µm. During this latter step, frequent monitoring of section thickness is essential. Occasionally, rotating a slide 180 degrees on the slide holder can be beneficial for improving the efficiency of grinding, but the technique should be used sparingly to avoid rounding of the section edges. If attrition rate is found to be non-uniform across a section (usually with larger samples), the positioning of the slide on the slide interface should be adjusted such that the thicker region of the section faces away from the centre of the abrasive wheel (where applicable). If this is predictable in advance (e.g. metal on one side), a piece of metal (same metal as the implant) can be embedded with the sample or later glued to the block on the side that is expected to have a higher attrition rate.

If sections are being prepared for backscatter electron microscopy, there is no need to measure or monitor thickness; simply perform the above steps to achieve a planar surface in preparation for polishing.

Polishing

Polishing requires the use of sequentially finer grits of sandpaper starting with p600. This should be done using the microgrinder in the oscillating mode while frequently monitoring thickness. Times should be optimized by trial and error and are generally kept very short (<10 min) to avoid over-polishing. As soon as near maximum polishing is achieved with each sandpaper grit, polishing should stop, and a finer sandpaper or diamond suspension should be used. I tend to go through an assort of p600, p800, p1200, p2500, and p4000 at 150–200 RPM before switching to polishing cloths and diamond suspensions (6 to 3 to 1 µm). The latter step is crucial if a sample is being processed for backscatter electron microscopy. A stereomicroscope can be used to monitor the progress of polishing and the quality of the final finish.

Conclusion

The design and associated protocol described here provides an inexpensive and practical solution to reproducibly prepare large biological samples containing metal implants for histological evaluations. Importantly, the high quality of sections prepared this way together with their resistance to deplasticization agents allows use of routine histological staining techniques with this method. The design provided can also be customized to retrofit any available wheel grinder/polisher. Timings should be optimized for each sample type.

Acknowledgments

I am grateful for the support provided by the Telstra Creator Space, Faculty of Engineering and Information Technology, The University of Melbourne, in the optimization and 3D printing of parts used in this project. Custom miniplate shown in Figure 2a was designed, manufactured and supplied by MAXONIQ, Melbourne, Australia. Figure 2b and all images in Figure 3 were acquired using a Zeiss Axioscan 7 as a part of a larger microscopy project (IMCRC Just-in-time patient-specific bone tumour project) with assistance from Kalyan Shobhana, Biological Optical Microscopy Platform, The University of Melbourne.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Supplementary material

Supplemental data for this article can be accessed online at https://doi.org/10.1080/01478885.2023.2205617

Additional information

Funding

This project was funded by the Department of Veterinary Biosciences, Melbourne Veterinary School, The University of Melbourne.