Ultrasonic bone cement removal efficiency in total joint arthroplasty revision: A computer tomographic‐based cadaver study

Polymethylmethacrylate (PMMA) removal during septic total joint arthroplasty revision is associated with a high fracture and perforation risk. Ultrasonic cement removal is considered a bone‐preserving technique. Currently, there is still a lack of sound data on efficacy as it is difficult to detect smaller residues with reasonable technical effort. However, incomplete removal is associated with the risk of biofilm coverage of the residue. Therefore, the study aimed to investigate the efficiency of ultrasonic‐based PMMA removal in a human cadaver model. The femoral components of a total hip and a total knee prosthesis were implanted in two cadaver femoral canals by 3rd generation cement fixation technique. Implants were then removed. Cement mantle extraction was performed with the OSCAR‐3‐System ultrasonic system (Orthofix®). Quantitative analysis of cement residues was carried out with dual‐energy and microcomputer tomography. With a 20 µm resolution, in vitro microcomputer tomography visualized tiniest PMMA residues. For clinical use, dual‐energy computer tomography tissue decomposition with 0.75 mm resolution is suitable. With ultrasound, more than 99% of PMMA was removed. Seven hundred thirty‐four residues with a mean volume of 0.40 ± 4.95 mm3 were identified with only 4 exceeding 1 cm in length in at least one axis. Ultrasonic cement removal of PMMA was almost complete and can therefore be considered a highly effective technique. For the first time, PMMA residues in the sub‐millimetre range were detected by computer tomography. Clinical implications of the small remaining PMMA fraction on the eradication rate of periprosthetic joint infection warrants further investigations.

tiniest PMMA residues. For clinical use, dual-energy computer tomography tissue decomposition with 0.75 mm resolution is suitable. With ultrasound, more than 99% of PMMA was removed. Seven hundred thirty-four residues with a mean volume of 0.40 ± 4.95 mm 3 were identified with only 4 exceeding 1 cm in length in at least one axis. Ultrasonic cement removal of PMMA was almost complete and can therefore be procedures are steadily increasing. 1 Consequently, it is logical to assume that the incidence of periprosthetic joint infections (PJI) will increase simultaneously. 2 PJI is a devastating complication of joint replacement surgery, accompanied by prolonged hospitalization, a high number of additional surgical interventions, limited joint mobility, and high mortality rates. 3,4 It is the third most common cause for revision in total hip arthroplasty (THA) and the most common reason for revision in total knee arthroplasty. 5 Polymethylmethacrylate (PMMA) bone cement is commonly removed using a combination of chisels, burrs, reverse cutting hooks, drills, and long-distance windowing. 10 Cement removal can be technically challenging, time-consuming and is associated with an increase in complications, for example, bleeding, bone damage, fractures, or perforations. [11][12][13] Cement removal methods based on ultrasound can be used to melt the cement mantle by heat generation. Ultrasound-powered tools have been used for decades with a low risk of cortical bone perforation. 14,15 However, there are only few studies on the use of ultrasound for cement removal in TJA revisions, which may partly be due to the difficulty of detecting minor cement residues with reasonable effort in a clinical context. Usually, the presence of cement residues is inspected visually or by fluoroscopy. Visible residues are removed mechanically-manually. Previous studies on ultrasound cement removal (UCR) did not quantitative assess its efficiency. We were interested precisely in this analysis at the level of small residues. Radiological methods offered themselves as imaging methods.
Therefore, another focus of our work was to determine the ability of different imaging modalities, that is, conventional, dualenergy CT (DECT) and micro-CT, to reliably detect small cement residues. These modalities, in the case of their suitability in terms of resolution and detection capability, were used to quantify the efficiency of UCR in TJA revision in a cadaver model.   performed in an idealized cylindrical polar geometry. To this end, an assumed central axis was constructed for each micro-CT section as the F I G U R E 1 Experimental procedure. The femoral neck and condyles were resected (A) and the femoral canal was prepared with rasps and drills (B). Once the correct implant size was achieved, the canal was irrigated with a pulsatile lavage (C). Cement was inserted under pressure (D) and both prostheses were implanted (E). Afterwards, the implants were removed and the cement extracted with ultrasound (F).

| Micro-CT imaging and analysis
line through the centre of mass of all non-air voxels which runs parallel to the scanner's longitudinal axis. Non-air voxels were identified by a thresholding procedure and included mostly bone and a few small fiducial markers placed on the sample exterior. Radial and azimuthal extents were calculated with respect to this axis, the latter first determined as an angular separation and subsequently converted to an arc length using the radius at the deposit's centre of mass. Longitudinal deposit extent was measured along the scan direction. All calculations were performed using a custom software built with SciPy libraries. 16 Measurements are reported as the difference between the overall minimum and maximum values across all voxels belonging to a deposit.
The maximum extent measured along any of the three directions was maintained per deposit and used to select deposits of sufficient size to be clinically relevant. This threshold was set at 10 mm.

| Micro-CT
Residual bone cement exhibited a heterogeneous, speckled texture, with Hounsfield units higher and lower than normal bone and slight streak artifacts (Figure 2). Cement deposits could not be segmented using a thresholding technique but had to be delineated manually ( Figure 6).
Overall, 99.4% of the PMMA was removed by UCR. The removal efficiency in terms of cement volume was estimated from the segmented deposits. Since micro-CT was not performed before cement removal, its original volume was estimated from the interior volume of the six bone segments after removal.

F I G U R E 3
Detection of bone cement by DECT. (A) A micro-CT slice is showing a prominent cement deposit identifiable through its speckled texture. (B, C) Coregistered CT image (120 kVp) and Zeff map. The cement deposit is not visible in the conventional 120 kVp-CT. In the DECT derived Rho/Z map for material decomposition, the cement deposit is clearly visible with a size extension comparable to that in the micro-CT. Contrast was adjusted to match the windowing used in Figure 8. DECT, dual-energy CT.

| DECT
While there is little contrast between cement and cortical bone in single spectrum CT imaging at 120 kVp, Z eff images calculated from dual-energy scans at 80 and 140 kVp allow satisfactory distinction between materials, that is, material decomposition ( Figure 8). Cement-bearing voxels were generally assigned an effective atomic number approximately one unit higher than surrounding cortical bone.

| DISCUSSION
Accompanied by an increase of TJA, revision surgery and PJI have gained in importance over the last years. 2   To understand the imaging results, it is important to recall that the PMMA-based bone cement used in this study contained 15% zirconium dioxide. Pure PMMA has a radiopacity similar to water and soft tissues. Only by adding material with a high electron density, for F I G U R E 6 Micro CT scan: (A) proximal section, (B) diaphyseal section, (C) distal Section 1: femur overview, 2: coronal view, 3: axial view, 4: sagittal view. example, zirconium, it is possible to differentiate between cement and bone using DECT. X-ray attenuation is determined by a material's electron density and its atomic number, as well as the energy of the X-ray photons. By acquiring images with two different energy spectra, it is possible to disentangle the former two material properties and distinguish materials which appear similar in singlespectrum imaging. Therefore, it seemed plausible that DECT 23,24 would lend itself very well to the task of differentiating bone tissue (A) The histogram shows the maximum extent of cement residues on one axis, (B) The histogram shows the volume of the residual cement deposits.

F I G U R E 8
CT imaging before and after cement extraction using effective atomic number (Zeff) and conventional CT number image (HU) and PMMA. Differentiation was indeed achieved using Z eff maps, mapping approximately the effective atomic number of materials.
Moreover, the PMMA residuals of the Z eff maps corresponded to the larger PMMA residuals of the micro-CT datasets. Therefore, it is justified to consider DECT as a well-suited candidate for a clinical method of identifying PMMA residuals, but its accuracy needs further detailed investigation.
Although ultrasonic tools have been used since the 1980s for cement removal, 25 15 Callaghan et al. demonstrated in an in vivo study that UCR has no disadvantageous effects on whole bone strength compared to manual cement removal. 32 Other histological studies showed no significant cortical bone damage. 33 Besides removing cement from the femoral canal, ultrasound was reported to extract a massive intrapelvic cement deposit 34 safely.
But care must be taken as ultrasonically driven tools can lead to thermal bone necrosis. These occur after exposure to 47°C for as little as 1 min 35 or within 30 s at 55°C. 36 Brooks et al. reported temperatures up to 80°C at the bone-cement interface and still temperatures above 40°C 60 s after the ultrasonic device was turned off. 37 Surrounding soft tissue can also be in severe danger of thermal necrosis. Goldberg et al. reported a case of complete proximal radial nerve palsy after UCR in the distal humerus when no saline irrigation was consistently used. 38 Thermal necrosis is also a possible complication of drill use. Studies have even shown less osteonecrosis and a more rapid bone healing using UCR compared to cement removal by high-speed burrs. 33 Effective temperature reduction during UCR is possible, for example, by using saline irrigation. 37 This study has several limitations: In-vivo study designs are subject to bias compared to clinical trials with patients. For instance, in our cadaver study without blood and surrounding soft tissues, we probably had a better degree of visualization than under clinical conditions. To obtain a reliable statement about the efficiency rates of the removal techniques, in a next step we want to compare a cohort of conventionally surgically treated patients with a cohort of ultrasound-guided patients in terms of removal efficiency using the dual-source technique. Besides, we did not use fluoroscopy during the cement removal to detect possible cement remnants. It is likely that if fluoroscopy had been used, we could have detected more PMMA residues.
Additionally, a sample size of two femurs for this investigation requires further studies with more samples. The challenging logistics within the study involving nonsurgical units while adhering to strict hygiene regulations did not allow for an expansion of cadaver studies at this point, so our principled study does not allow for a statistically sound statement in general. Another limitation of this study is the fact that we cannot exclude an impact of the surgeon's skill level on the cement removal efficiency. Furthermore, an analysis of cortical bone loss during cement removal was not conducted. Micro-CT cement residues were manually segmented. While large residues were easily identified, sensitivity to the smallest residues was limited with an unquantifiable number not segmented. Additionally, small residues were sometimes difficult to distinguish from the bone and some could not be fully segmented, as they were located close to the scan edges.
In addition, some residues (n = 20) extended beyond the scanning range of the sections. Therefore, they could not be fully segmented, and their volume and extents were underestimated. Differences in Z eff between bone cement and bone could potentially be used for automated residual cement detection. It has to be noted that erroneously elevated Z eff values have also been found in artifacts at bone-air interfaces that might be mistaken for cement. For automated detection these artifacts have to be suppressed which requires a detailed understanding of the underlying algorithms for Z eff derivation.
The potential benefits of DECT must also be weighed against the additional patient dose required to use this technique.
Nevertheless, in this study we achieved a quantitative assessment regarding the efficiency of cement removal. To this end, we first used ex vivo micro-CT as an almost infallible method for detecting tiny cement residues, but one that is limited to small samples. For clinical application, we used the dual-energy-based technique that also promises valid detection of small cement residues. Thus, both CT-based imaging modalities are able to reliably visualise cement residues and demonstrate for the first time that UCR leads to an almost complete removal of bone cement.

| CONCLUSION
UCR has proven to be a highly effective technique for almost complete and safe removal of bone cement. However, dual-energy and micro-CT techniques have been able to visualize the presence of PMMA residues in the millimetre and submillimeter range, with the former technique being applicable in a clinical setting. Further studies need to be conducted to assess the clinical impact of these cement residuals.