- Email: [email protected]
Laser Scanning as a Useful Tool in Implant Retrieval Analysis: A Demonstration Using Rotating Platform and Fixed Bearing Tibial Inserts
The Journal of Arthroplasty 28 Suppl. 1 (2013) 152–156
Contents lists available at ScienceDirect
The Journal of Arthroplasty journal homepage: www.arthroplastyjournal.org
Laser Scanning as a Useful Tool in Implant Retrieval Analysis: A Demonstration Using Rotating Platform and Fixed Bearing Tibial Inserts Kirsten E. Stoner, MEng a, Nader A. Nassif, MD b, Timothy M. Wright, PhD a, Douglas E. Padgett, MD b a b
Department of Orthopaedic Surgery, Hospital for Special Surgery, New York, New York Department of Biomechanics, Hospital for Special Surgery, New York, New York
a r t i c l e
i n f o
Article history: Received 15 October 2012 Accepted 2 July 2013 Keywords: laser scanning arthroplasty implant retrieval
a b s t r a c t Objective methods for analyzing arthroplasty retrieval implants are needed. To address this, we used a readily available laser scanner to analyze damage deviations between cohorts of rotating platform and fixed bearing inserts previously analyzed using traditional, subjective retrieval analysis methods. We asked the following research questions: 1) Do articular surface deviations measured by the scanner correlate with the subjective damage scores? 2) Do articular surface deviations differ between inserts due to design differences? Correlations between deviations and damage scores were present in RP but not FB inserts. Seven different deviation patterns were present between the RP and FB inserts and were a function of design. In conclusion laser scanning was found to be a useful objective tool for analyzing arthroplasty retrievals. © 2013 Elsevier Inc. All rights reserved.
Throughout the development of modern total knee arthroplasty (TKA), retrieval analysis has been used to assess the in vivo performance of the implant components [1,2]. For example, retrieval analysis enabled us to determine tibial post performance leading to improved post designs for posterior stabilized TKAs [3]. Additionally, retrieval analysis showed problems in polyethylene quality [4] and how polyethylene oxidation contributed to wear damage. [5] Finally, similarities between in vivo damage and in vitro laboratory studies were used to validate the usefulness of such preclinical experiments [6]. Damage in retrieved implants has historically been assessed subjectively [1]. Recently, investigators have developed more objective measurements such as damage mapping to measure areas covered by specific damage modes [2] and caliper measurements to determine localized polyethylene thickness changes [7,8]. However, damage mapping is at best semi-quantitative, and caliper measurements do not easily provide thickness measurements over an entire surface. Microcomputed tomography can provide 3D images of retrieved polyethylene inserts [9–11], providing accurate, high resolution data, but scanning is time consuming, is costly, and requires specialized training and safety precautions. In contrast, quantitative optical imaging systems are considerably cheaper, easier to use, and safer, and provide much quicker scan times (typically less than an hour). We examined the effectiveness of using an inexpensive, commercial tabletop laser scanner in objectively measuring articular surface We thank our funding sources: The Mary and Fred Trump Institute for Retrieval Analysis and the Clark and Kirby Foundations. The Conflict of Interest statement associated with this article can be found at http:// dx.doi.org/10.1016/j.arth.2013.07.004. Reprint requests: Timothy Wright, Ph.D., HSS Department of Biomechanics, 535 East 70th Street, New York, NY 10021. 0883-5403/2808-0037$36.00/0 – see front matter © 2013 Elsevier Inc. All rights reserved. http://dx.doi.org/10.1016/j.arth.2013.07.004
damage and geometrical deviations of retrieved polyethylene TKA tibial inserts. We compared two types of inserts, rotating platform (RP) and fixed bearing (FB), that had been previously been investigated in our lab [2,12], because differences in these designs had already been established with traditional, subjective scoring methods. We asked the following research questions: 1) Do articular surface deviations measured by the scanner correlate with the subjective damage scores? 2) Do articular surface deviations differ between inserts due to design differences? Materials and Methods We identified RP and FB tibial inserts from one TKA system from one orthopaedic manufacturer (Posterior Stabilized (STB) Sigma, Depuy, Inc, Warsaw, IN) from our IRB-approved institutional retrieved implant system. The RP and FB versions utilize the same femoral component, but the tibial articular geometry differs; the RP is more conforming with the femoral component than is the FB insert. Because FB post design in this TKA system changed in 1998, FB inserts that were implanted prior to 1998 were excluded. All of the inserts had been previously graded for damage in an earlier study [2,12]: 20 RP and 13 FB inserts, all fabricated from 1020 GUR ultrahigh molecular weight polyethylene resin and all gamma sterilized in air-impermeable foil prior to implantation. The FB inserts had been implanted between 1998 and 2010; the RP inserts had been implanted between 2001 and 2007. The RP cohort consisted of the following insert sizes: 2 (n = 1), 2.5 (n = 4), 3 (n = 6), 4 (n = 5), and 5 (n = 4). The FB cohort had inserts of size 2.5 (n = 1), 3 (n = 5), 4 (n = 5), 5 (n = 2). Patient charts were queried for gender, BMI, age at time of revision, length of implantation (LOI), and reason for revision
K.E. Stoner et al. / The Journal of Arthroplasty 28 Suppl. 1 (2013) 152–156
(Table 1). Component alignment in the coronal and sagittal planes was measured by one author (NN) from the coronal and lateral prerevision radiographs, respectively, for all inserts except one FB insert, which had been revised for a periprosthetic fracture, precluding alignment measurements. Clinical data were similar between the patients who had received the RP and FB inserts, except for LOI and coronal tibial and femoral alignment. The mean LOI for RP inserts was significantly (P = 0.028) shorter (2.1 years, ranging from 0.4 to 3.8 years) than that for the FB inserts (6.7 years, ranging from 0.4 to 12.5 years). RP inserts were aligned in significantly (P = 0.039) less varus with a coronal tibial angle of 1.5° ± 3.6° varus vs. 2.9° ± 3.1° varus for the FB inserts. Similarly, RP inserts had an average coronal femoral angle of 7.0° ± 3.7° valgus, significantly greater than the 4.3° ± 3.2° valgus for the FB inserts (P = 0.046). Subjective damage scores had been previously assigned [2,12] independently by two observers to the articular surfaces of the tibial inserts using the Hood grading system [1]. Briefly, the articular surface of the inserts was separated into 10 regions, including the tibial plateaus and the post. Each region was graded for seven modes of damage: burnishing, scratching, pitting, delamination, abrasion, embedded debris, and deformation and given a score of 0–3 where higher scores indicate more severe damage. After scoring, the articular surfaces of all inserts were digitized using a commercially available desktop 3D scanner (NextEngine, Santa Monica CA). The scanner uses laser reflection technology and produces a point cloud with a resolution of 63 microns [12]. Twelve individual scans were taken of the insert to ensure that the entire surface was imaged. Six of the scans were taken at 0° of inclination with a scan taken at every 30° of revolution. Three more scans were taken with the insert at an inclination of 45° with a scan taken again at every 30° of revolution. Three final scans were taken with the insert at an inclination of − 35° with a scan taken every 30° of revolution. The entire scan time was ~ 40 min/insert. The 12 scans were then aligned and combined using the manufacturer’s software (NextEngine ScanStudio Ver 1.3.2). The resulting combined model was a point cloud of ~ 300,000 points. Pristine inserts of each of the sizes that were included in the RP and FB retrieved inserts were purchased from the manufacturer, and scanned in the same manner. The 3D models from the scans of the retrieved inserts were compared to those from the size-matched pristine models using
153
Geomagic Qualify software (Ver. 12.0, Morrisville, NC). The articular surface of the retrieved insert was aligned to that of the pristine articular surface using the “Best Fit Alignment” tool. Briefly, an iterative closest point algorithm was applied using a least-squares method of minimizing the sums of squares distances between the scans and then realigning them. Deviations in dimensions between the retrieved and pristine inserts were then depicted colorimetrically. We used the term “deviation” in describing the geometric differences between the surfaces because we do not know if these differences were due to wear (removal of material from the surface), deformation (due to yielding or creep of the polyethylene), or some combination of the two. The 3D models depicting the articular surfaces were further sectioned into four regions of interest: medial plateau, lateral plateau, anterior post, and posterior post (Fig. 1), and deviation values were averaged for each region. To determine regional deviation rate, the average deviation for the region was divided by the LOI for the insert. Differences between damage scores, demographic and radiographic variables, and deviations for the RP and FB inserts were determined by a t-test, for non-parametric data a Mann–Whitney Rank Sum test was performed. Differences in the deviations between plateau regions and between post regions within each design were also determined by a t-test, for non-parametric data a Mann–Whitney Rank Sum test was performed. Correlations between demographic variables and both damage scores and deviations were examined using linear regression analysis. P b 0.05 was considered significant. Results Subjective damage scoring revealed regional differences between RP and FB inserts. The RP inserts had significantly higher damages scores on both the medial and lateral tibial plateaus than the FB inserts (medially, 18 ± 5 for the RP and 14 ± 5, P = 0.022; laterally, 18 ± 5 vs. 13 ± 6, P = 0.012). Conversely, the damage scores for the post regions were significantly (P b 0.001) lower in the RP (0.2 ± 0.8) than the FB inserts (3.2 ± 2.1). No differences in regional deviations were seen between RP and FB inserts (Fig. 2); however, RP inserts had significantly greater medial and lateral deviation rates (P = 0.016 and P = 0.028, respectively, Fig. 3). No differences in deviations were seen within a design
Table 1 Demographic and Radiographic Data of Rotating Platform and Fixed Bearing Retrieved Cohorts. Rotating Platform (n = 20) Age (years) Gender (M:F) BMI Length of Implantation (years) Reasons for Revision (#)
Radiographic Analysis (°) Coronal Femoral Tibial Femoral Tibial Lateral Femoral Tibial
Fixed Bearing (n = 13)
66 ± 11 8:12 29 ± 5 2.1 ± 1.0
68 ± 13 12:7 31 ± 6 6.7 ± 4.6 (P = 0.028)
Infection (6) Stiffness (7) Osteolysis/Loosening (4) Instability (2) Malpositioned Component (1) Fracture (0)
Infection (4) Stiffness (2) Osteolysis/Loosening (6) Instability (0) Malpositioned Component (0) Fracture (1)
5.6 ± 5.5 7.0 ± 3.7 −1.5 ± 3.6
2.8 ± 5.8 4.3 ± 3.2 (P = 0.046) −2.9 ± 3.1 (P = 0.039)
3.0 ± 4.4 3.2 ± 4.8
2.9 ± 2.1 6.0 ± 6.0
Data are presented as mean and standard deviations. For coronal radiographic measurements negative values imply varus and positive valgus. In lateral measurements negative values imply extension and positive values flexion.
Fig. 1. The articular surface data were considered for four separate regions: the medial plateau region, the lateral plateau region, the posterior post, and the anterior post.
154
K.E. Stoner et al. / The Journal of Arthroplasty 28 Suppl. 1 (2013) 152–156
between the medial and lateral plateaus or between anterior and posterior sides of the post. A positive correlation was found between articular surface damage scores and surface deviations in RP inserts for both the lateral and medial plateaus (P = 0.010 and 0.004, respectively). However, no correlation existed between articular surface damage scores and deviations in FB inserts (Fig. 4). Additionally, no correlations were found between BMI, LOI, age, and alignment with damage scores or deviations. Interestingly, seven distinct patterns were identified from the colorimetric deviation maps (Fig. 5). Common to both RP and FB designs were retrieved inserts with little to no deviation (4 of 20 RP and 2 of 13 FB inserts). RP inserts exhibited four patterns: posterior (in 8 of 20) with deviations located towards the posterior medial and lateral plateaus (Fig. 5A); middle stripe (in 1 of 20) with deviations located down the midline of the plateaus (Fig. 5B); a combination of the first two patterns (in 3 of 20) with deviations located both posteriorly and down the midline of the plateaus (Fig. 5C), and an entire surface pattern (in 4 of 20) with deviations equally encompassing both plateau surfaces (Fig. 5D). Conversely, FB inserts exhibited two patterns: peripheral (in 2 of 13) with deviations located along the outside edge of the plateaus (Fig. 5E) and asymmetric (in 9 of 13) with deviations asymmetrically distributed between the medial and lateral plateaus (Fig. 5F). Unlike the articular surfaces, the anterior and posterior surfaces of the posts had a common pattern. Both RP and FB posts showed primarily symmetric contact on the posterior post, consistent with contact against the cam of the femoral component. Five of the 13 FB inserts showed an anterior “bowtie” pattern consistent with unintended articulation with the anterior intercondylar box of the femoral component. Discussion
Fig. 3. Average deviation rates of medial and lateral condylar surfaces are shown for rotating platform and fixed bearing articular surfaces.
subjective damage scores? 2) Do articular surface deviations differ between inserts due to design differences? Regarding our first question, higher damage scores correlated with higher deviations in rotating platform inserts, but not in fixed bearing inserts. This discrepancy may in fact not reflect differences caused by design, but rather results from the shorter implantation times in the RP inserts. The shorter length of implantation may reflect that these inserts were still “bedding in,” a phenomenon common in total hip arthroplasty in which the femoral head deforms
Retrieval analysis has been used to assess the performance of orthopaedic devices for decades. However, the need remains for ever more objective methods of assessing damage as useful technologies emerge. 3D registration and modeling of retrieved implants through techniques such as microcomputed tomography have proved promising [9,10]; however, the technique can be expensive and may not readily be available to most investigators. Alternatively, commercially available table top laser scanners are less expensive and easier to use. We used such a scanner to analyze retrieved TKA inserts of two established designs to answer the following questions: 1) Do articular surface deviations measured by the laser scanner correlate with
Fig. 2. Average deviations of medial and lateral condylar area are shown for rotating platform and fixed bearing articular surfaces.
Fig. 4. Correlations are shown between medial and lateral damage scores and medial and lateral deviations, respectively, for rotating platform (A) and fixed bearing (B) retrieved inserts.
K.E. Stoner et al. / The Journal of Arthroplasty 28 Suppl. 1 (2013) 152–156
155
Fig. 5. Four deviation patterns were found for the rotating platform inserts (A: Posterior, B: Middle Stripe, C: Combination, D: Entire Surface) and two different patterns were found for the fixed-bearing inserts (E: Peripheral, F: Asymmetric). Units are in mm. Blue areas indicate deviations below the original surface.
the polyethylene of the acetabular component in the initial postoperative period. Perhaps this is reflected in TKAs by the correlation of damage scores with deviations only during the initial time after implantation. A wear simulation study reported that the deformation rate of fixed bearing tibial inserts did indeed decrease with time [13]. In an earlier study of residual stresses created in knee-like non-conforming surfaces under cyclic loading, we found convergence to a steady-state cycle of stress and deformation after only a few cycles [14]. Another possible reason for the difference in deviations between RP and FB inserts is reflected in the deviation patterns that we found in these two cohorts. Rotating platforms had more symmetric deviations across both plateaus compared to the fixed bearing inserts. Averaging the deviation over each plateau may have skewed the data for asymmetric deviation patterns such as were found for the FB inserts, thus negatively affecting the correlation between damage score and deviation. This would suggest that these correlations might be strengthened by examining the correlations across many, much smaller regions of the articular surfaces. We are currently investigating the use of image stitching techniques to perform such correlations on a pixel-by-pixel basis between digital photographs of the articular surfaces (rather than subjective scores) and deviation measurements. As to our second research question, the deviation patterns did vary between the RP and FB inserts. The symmetric distribution of deviations between the medial and lateral plateaus of the RP design reflects the higher conformity of the tibiofemoral articulation coupled with the rotation allowed at the insert–tray interface. This motion allows both femoral condyles to be in equal contact with the polyethylene insert. The lack of anterior post “bowtie” damage in the RP inserts is consistent with our previous study of a larger cohort of RP versus FB inserts [2,12], and demonstrates that less impingement with the intercondylar box of the femur occurs when the insert is allowed to rotate. The more axisymmetric deviation patterns found in the FB inserts likely reflect the lower tibiofemoral conformity and the fact that the insert is locked in the tray. These features allow the femur to slide across the polyethylene insert. The higher deviations on the medial plateau are consistent with higher loads being transferred through the
medial plateau, as has been shown with loads measured from instrumented TKAs [15]. Additionally the “bowtie” pattern present on many of the FB tibial posts indicate that unlike the RP, the femoral component impinges with the post, affecting the translation and rotation of the femoral relative to the tibial component. Our study has several limitations. First, it is a retrieval study, so our evaluation was limited to failed implants, many of which were revised for reasons (such as stiffness, osteolysis, and loosening) associated with early failure and abnormal wear. Second our cohort is small, which may have limited our ability to determine differences. Third, the original dimensions of the inserts were unknown, so we relied on comparison to pristine inserts of the same design and manufacturer. Nonetheless, this precluded us from determining the absolute volumetric wear even assuming manufacturing tolerances [9]. Finally, the 3D laser scanning system measures the deformation in the retrieved components; however; it cannot distinguish deviations due from plastic deformation from those caused by material loss through wear. In summary, the commercially available table top scanner could measure differences in deviations between two similar, but different implant designs. Most importantly, with the scanner we could obtain subjective data over a larger area, allowing us to determine differences in in vivo performance between RP and FB implants. This coupled with the availability, ease of use, speed, and low cost makes it a useful tool for modern retrieval analysis. Acknowledgment We thank Marcella Elpers for her assistance in implant retrieval. We also thank Seth Jerabek, MD, Stephanie Tow, and Rose Fu for assistance with implant grading. References 1. Hood RW, Wright TM, Burstein AH. Retrieval analysis of total knee prostheses: a method and its application to 48 total condylar prostheses. J Biomed Mater Res 1983;17:829. 2. Kelly NH, Fu RH, Wright TM, et al. Wear damage in mobile-bearing TKA is as severe as that in fixed-bearing TKA. Clin Orthop Relat Res 2011;469:123.
156
K.E. Stoner et al. / The Journal of Arthroplasty 28 Suppl. 1 (2013) 152–156
3. Dolan MM, Kelly NH, Nguyen JT, et al. Implant design influences tibial post wear damage in posterior-stabilized Knees. Clin Orthop 2011;469:160. 4. Tsao A, Mintz L, McRae C, et al. Failure of the porous-coated anatomic prosthesis in total knee arthroplasty due to severe polyethylene wear. J Bone Joint Surg 1993;75:19. 5. Medel FJ, Kurtz SM, Parvizi J, et al. In vivo oxidation contributes to delamination but not pitting in polyethylene components for total knee arthroplasty. J Arthroplasty 2011;26:802. 6. Rawlinson JJ, Furman BD, Li S, et al. Retrieval, experimental, and computational assessment of the performance of total knee replacements. J Orthop Res 2006; 24:1384. 7. Engh GA, Zimmerman RL, Parks NL, et al. Analysis of wear in retrieved mobile and fixed bearing knee inserts. J Arthroplasty 2009;24:28. 8. Berry DJ, Currier JH, Mayor MB, et al. Knee wear measured in retrievals: a polished tray reduces insert wear. Clin Orthop Relat Res 2012;470:1860. 9. Teeter MG, Douglas D, Naudie R, et al. In vitro quantification of wear in tibial inserts using microcomputed tomography. Clin Orthop Relat Res 2011;469:107.
10. Teeter MG, Naudie DDR, Milner JS, et al. Determination of reference geometry for polyethylene tibial insert wear analysis. J Arthroplasty 2011;26:497. 11. Engh CA, Zimmer RL, Hopper RH, et al. Can microcomputed tomography measure retrieved polyethylene wear? Comparing fixed-bearing and rotating-platform knees. Clin Orthop Relat Res 2013;471:86. 12. Stoner K, Jerabek SA, Tow S, et al. Rotating-platform has no surface damage advantage over fixed-bearing TKA. Clin Orthop Relat Res 2013;471:76. 13. Harman MK, DesJardins J, Benson L, et al. Comparison of polyethylene tibial insert damage from in vivo function and in vitro wear simulation. J Orthop Res Off Publ Orthop Res Soc 2009;27:540. 14. Estupiñán JA, Bartel DL, Wright TM. Residual stresses in ultra-high molecular weight polyethylene loaded cyclically by a rigid moving indenter in nonconforming geometries. J Orthop Res 1998;16:80. 15. Mündermann A, Dyrby CO, D’Lima DD, et al. In vivo knee loading characteristics during activities of daily living as measured by an instrumented total knee replacement. J Orthop Res 2008;26:1167.
Contents lists available at ScienceDirect
The Journal of Arthroplasty journal homepage: www.arthroplastyjournal.org
Laser Scanning as a Useful Tool in Implant Retrieval Analysis: A Demonstration Using Rotating Platform and Fixed Bearing Tibial Inserts Kirsten E. Stoner, MEng a, Nader A. Nassif, MD b, Timothy M. Wright, PhD a, Douglas E. Padgett, MD b a b
Department of Orthopaedic Surgery, Hospital for Special Surgery, New York, New York Department of Biomechanics, Hospital for Special Surgery, New York, New York
a r t i c l e
i n f o
Article history: Received 15 October 2012 Accepted 2 July 2013 Keywords: laser scanning arthroplasty implant retrieval
a b s t r a c t Objective methods for analyzing arthroplasty retrieval implants are needed. To address this, we used a readily available laser scanner to analyze damage deviations between cohorts of rotating platform and fixed bearing inserts previously analyzed using traditional, subjective retrieval analysis methods. We asked the following research questions: 1) Do articular surface deviations measured by the scanner correlate with the subjective damage scores? 2) Do articular surface deviations differ between inserts due to design differences? Correlations between deviations and damage scores were present in RP but not FB inserts. Seven different deviation patterns were present between the RP and FB inserts and were a function of design. In conclusion laser scanning was found to be a useful objective tool for analyzing arthroplasty retrievals. © 2013 Elsevier Inc. All rights reserved.
Throughout the development of modern total knee arthroplasty (TKA), retrieval analysis has been used to assess the in vivo performance of the implant components [1,2]. For example, retrieval analysis enabled us to determine tibial post performance leading to improved post designs for posterior stabilized TKAs [3]. Additionally, retrieval analysis showed problems in polyethylene quality [4] and how polyethylene oxidation contributed to wear damage. [5] Finally, similarities between in vivo damage and in vitro laboratory studies were used to validate the usefulness of such preclinical experiments [6]. Damage in retrieved implants has historically been assessed subjectively [1]. Recently, investigators have developed more objective measurements such as damage mapping to measure areas covered by specific damage modes [2] and caliper measurements to determine localized polyethylene thickness changes [7,8]. However, damage mapping is at best semi-quantitative, and caliper measurements do not easily provide thickness measurements over an entire surface. Microcomputed tomography can provide 3D images of retrieved polyethylene inserts [9–11], providing accurate, high resolution data, but scanning is time consuming, is costly, and requires specialized training and safety precautions. In contrast, quantitative optical imaging systems are considerably cheaper, easier to use, and safer, and provide much quicker scan times (typically less than an hour). We examined the effectiveness of using an inexpensive, commercial tabletop laser scanner in objectively measuring articular surface We thank our funding sources: The Mary and Fred Trump Institute for Retrieval Analysis and the Clark and Kirby Foundations. The Conflict of Interest statement associated with this article can be found at http:// dx.doi.org/10.1016/j.arth.2013.07.004. Reprint requests: Timothy Wright, Ph.D., HSS Department of Biomechanics, 535 East 70th Street, New York, NY 10021. 0883-5403/2808-0037$36.00/0 – see front matter © 2013 Elsevier Inc. All rights reserved. http://dx.doi.org/10.1016/j.arth.2013.07.004
damage and geometrical deviations of retrieved polyethylene TKA tibial inserts. We compared two types of inserts, rotating platform (RP) and fixed bearing (FB), that had been previously been investigated in our lab [2,12], because differences in these designs had already been established with traditional, subjective scoring methods. We asked the following research questions: 1) Do articular surface deviations measured by the scanner correlate with the subjective damage scores? 2) Do articular surface deviations differ between inserts due to design differences? Materials and Methods We identified RP and FB tibial inserts from one TKA system from one orthopaedic manufacturer (Posterior Stabilized (STB) Sigma, Depuy, Inc, Warsaw, IN) from our IRB-approved institutional retrieved implant system. The RP and FB versions utilize the same femoral component, but the tibial articular geometry differs; the RP is more conforming with the femoral component than is the FB insert. Because FB post design in this TKA system changed in 1998, FB inserts that were implanted prior to 1998 were excluded. All of the inserts had been previously graded for damage in an earlier study [2,12]: 20 RP and 13 FB inserts, all fabricated from 1020 GUR ultrahigh molecular weight polyethylene resin and all gamma sterilized in air-impermeable foil prior to implantation. The FB inserts had been implanted between 1998 and 2010; the RP inserts had been implanted between 2001 and 2007. The RP cohort consisted of the following insert sizes: 2 (n = 1), 2.5 (n = 4), 3 (n = 6), 4 (n = 5), and 5 (n = 4). The FB cohort had inserts of size 2.5 (n = 1), 3 (n = 5), 4 (n = 5), 5 (n = 2). Patient charts were queried for gender, BMI, age at time of revision, length of implantation (LOI), and reason for revision
K.E. Stoner et al. / The Journal of Arthroplasty 28 Suppl. 1 (2013) 152–156
(Table 1). Component alignment in the coronal and sagittal planes was measured by one author (NN) from the coronal and lateral prerevision radiographs, respectively, for all inserts except one FB insert, which had been revised for a periprosthetic fracture, precluding alignment measurements. Clinical data were similar between the patients who had received the RP and FB inserts, except for LOI and coronal tibial and femoral alignment. The mean LOI for RP inserts was significantly (P = 0.028) shorter (2.1 years, ranging from 0.4 to 3.8 years) than that for the FB inserts (6.7 years, ranging from 0.4 to 12.5 years). RP inserts were aligned in significantly (P = 0.039) less varus with a coronal tibial angle of 1.5° ± 3.6° varus vs. 2.9° ± 3.1° varus for the FB inserts. Similarly, RP inserts had an average coronal femoral angle of 7.0° ± 3.7° valgus, significantly greater than the 4.3° ± 3.2° valgus for the FB inserts (P = 0.046). Subjective damage scores had been previously assigned [2,12] independently by two observers to the articular surfaces of the tibial inserts using the Hood grading system [1]. Briefly, the articular surface of the inserts was separated into 10 regions, including the tibial plateaus and the post. Each region was graded for seven modes of damage: burnishing, scratching, pitting, delamination, abrasion, embedded debris, and deformation and given a score of 0–3 where higher scores indicate more severe damage. After scoring, the articular surfaces of all inserts were digitized using a commercially available desktop 3D scanner (NextEngine, Santa Monica CA). The scanner uses laser reflection technology and produces a point cloud with a resolution of 63 microns [12]. Twelve individual scans were taken of the insert to ensure that the entire surface was imaged. Six of the scans were taken at 0° of inclination with a scan taken at every 30° of revolution. Three more scans were taken with the insert at an inclination of 45° with a scan taken again at every 30° of revolution. Three final scans were taken with the insert at an inclination of − 35° with a scan taken every 30° of revolution. The entire scan time was ~ 40 min/insert. The 12 scans were then aligned and combined using the manufacturer’s software (NextEngine ScanStudio Ver 1.3.2). The resulting combined model was a point cloud of ~ 300,000 points. Pristine inserts of each of the sizes that were included in the RP and FB retrieved inserts were purchased from the manufacturer, and scanned in the same manner. The 3D models from the scans of the retrieved inserts were compared to those from the size-matched pristine models using
153
Geomagic Qualify software (Ver. 12.0, Morrisville, NC). The articular surface of the retrieved insert was aligned to that of the pristine articular surface using the “Best Fit Alignment” tool. Briefly, an iterative closest point algorithm was applied using a least-squares method of minimizing the sums of squares distances between the scans and then realigning them. Deviations in dimensions between the retrieved and pristine inserts were then depicted colorimetrically. We used the term “deviation” in describing the geometric differences between the surfaces because we do not know if these differences were due to wear (removal of material from the surface), deformation (due to yielding or creep of the polyethylene), or some combination of the two. The 3D models depicting the articular surfaces were further sectioned into four regions of interest: medial plateau, lateral plateau, anterior post, and posterior post (Fig. 1), and deviation values were averaged for each region. To determine regional deviation rate, the average deviation for the region was divided by the LOI for the insert. Differences between damage scores, demographic and radiographic variables, and deviations for the RP and FB inserts were determined by a t-test, for non-parametric data a Mann–Whitney Rank Sum test was performed. Differences in the deviations between plateau regions and between post regions within each design were also determined by a t-test, for non-parametric data a Mann–Whitney Rank Sum test was performed. Correlations between demographic variables and both damage scores and deviations were examined using linear regression analysis. P b 0.05 was considered significant. Results Subjective damage scoring revealed regional differences between RP and FB inserts. The RP inserts had significantly higher damages scores on both the medial and lateral tibial plateaus than the FB inserts (medially, 18 ± 5 for the RP and 14 ± 5, P = 0.022; laterally, 18 ± 5 vs. 13 ± 6, P = 0.012). Conversely, the damage scores for the post regions were significantly (P b 0.001) lower in the RP (0.2 ± 0.8) than the FB inserts (3.2 ± 2.1). No differences in regional deviations were seen between RP and FB inserts (Fig. 2); however, RP inserts had significantly greater medial and lateral deviation rates (P = 0.016 and P = 0.028, respectively, Fig. 3). No differences in deviations were seen within a design
Table 1 Demographic and Radiographic Data of Rotating Platform and Fixed Bearing Retrieved Cohorts. Rotating Platform (n = 20) Age (years) Gender (M:F) BMI Length of Implantation (years) Reasons for Revision (#)
Radiographic Analysis (°) Coronal Femoral Tibial Femoral Tibial Lateral Femoral Tibial
Fixed Bearing (n = 13)
66 ± 11 8:12 29 ± 5 2.1 ± 1.0
68 ± 13 12:7 31 ± 6 6.7 ± 4.6 (P = 0.028)
Infection (6) Stiffness (7) Osteolysis/Loosening (4) Instability (2) Malpositioned Component (1) Fracture (0)
Infection (4) Stiffness (2) Osteolysis/Loosening (6) Instability (0) Malpositioned Component (0) Fracture (1)
5.6 ± 5.5 7.0 ± 3.7 −1.5 ± 3.6
2.8 ± 5.8 4.3 ± 3.2 (P = 0.046) −2.9 ± 3.1 (P = 0.039)
3.0 ± 4.4 3.2 ± 4.8
2.9 ± 2.1 6.0 ± 6.0
Data are presented as mean and standard deviations. For coronal radiographic measurements negative values imply varus and positive valgus. In lateral measurements negative values imply extension and positive values flexion.
Fig. 1. The articular surface data were considered for four separate regions: the medial plateau region, the lateral plateau region, the posterior post, and the anterior post.
154
K.E. Stoner et al. / The Journal of Arthroplasty 28 Suppl. 1 (2013) 152–156
between the medial and lateral plateaus or between anterior and posterior sides of the post. A positive correlation was found between articular surface damage scores and surface deviations in RP inserts for both the lateral and medial plateaus (P = 0.010 and 0.004, respectively). However, no correlation existed between articular surface damage scores and deviations in FB inserts (Fig. 4). Additionally, no correlations were found between BMI, LOI, age, and alignment with damage scores or deviations. Interestingly, seven distinct patterns were identified from the colorimetric deviation maps (Fig. 5). Common to both RP and FB designs were retrieved inserts with little to no deviation (4 of 20 RP and 2 of 13 FB inserts). RP inserts exhibited four patterns: posterior (in 8 of 20) with deviations located towards the posterior medial and lateral plateaus (Fig. 5A); middle stripe (in 1 of 20) with deviations located down the midline of the plateaus (Fig. 5B); a combination of the first two patterns (in 3 of 20) with deviations located both posteriorly and down the midline of the plateaus (Fig. 5C), and an entire surface pattern (in 4 of 20) with deviations equally encompassing both plateau surfaces (Fig. 5D). Conversely, FB inserts exhibited two patterns: peripheral (in 2 of 13) with deviations located along the outside edge of the plateaus (Fig. 5E) and asymmetric (in 9 of 13) with deviations asymmetrically distributed between the medial and lateral plateaus (Fig. 5F). Unlike the articular surfaces, the anterior and posterior surfaces of the posts had a common pattern. Both RP and FB posts showed primarily symmetric contact on the posterior post, consistent with contact against the cam of the femoral component. Five of the 13 FB inserts showed an anterior “bowtie” pattern consistent with unintended articulation with the anterior intercondylar box of the femoral component. Discussion
Fig. 3. Average deviation rates of medial and lateral condylar surfaces are shown for rotating platform and fixed bearing articular surfaces.
subjective damage scores? 2) Do articular surface deviations differ between inserts due to design differences? Regarding our first question, higher damage scores correlated with higher deviations in rotating platform inserts, but not in fixed bearing inserts. This discrepancy may in fact not reflect differences caused by design, but rather results from the shorter implantation times in the RP inserts. The shorter length of implantation may reflect that these inserts were still “bedding in,” a phenomenon common in total hip arthroplasty in which the femoral head deforms
Retrieval analysis has been used to assess the performance of orthopaedic devices for decades. However, the need remains for ever more objective methods of assessing damage as useful technologies emerge. 3D registration and modeling of retrieved implants through techniques such as microcomputed tomography have proved promising [9,10]; however, the technique can be expensive and may not readily be available to most investigators. Alternatively, commercially available table top laser scanners are less expensive and easier to use. We used such a scanner to analyze retrieved TKA inserts of two established designs to answer the following questions: 1) Do articular surface deviations measured by the laser scanner correlate with
Fig. 2. Average deviations of medial and lateral condylar area are shown for rotating platform and fixed bearing articular surfaces.
Fig. 4. Correlations are shown between medial and lateral damage scores and medial and lateral deviations, respectively, for rotating platform (A) and fixed bearing (B) retrieved inserts.
K.E. Stoner et al. / The Journal of Arthroplasty 28 Suppl. 1 (2013) 152–156
155
Fig. 5. Four deviation patterns were found for the rotating platform inserts (A: Posterior, B: Middle Stripe, C: Combination, D: Entire Surface) and two different patterns were found for the fixed-bearing inserts (E: Peripheral, F: Asymmetric). Units are in mm. Blue areas indicate deviations below the original surface.
the polyethylene of the acetabular component in the initial postoperative period. Perhaps this is reflected in TKAs by the correlation of damage scores with deviations only during the initial time after implantation. A wear simulation study reported that the deformation rate of fixed bearing tibial inserts did indeed decrease with time [13]. In an earlier study of residual stresses created in knee-like non-conforming surfaces under cyclic loading, we found convergence to a steady-state cycle of stress and deformation after only a few cycles [14]. Another possible reason for the difference in deviations between RP and FB inserts is reflected in the deviation patterns that we found in these two cohorts. Rotating platforms had more symmetric deviations across both plateaus compared to the fixed bearing inserts. Averaging the deviation over each plateau may have skewed the data for asymmetric deviation patterns such as were found for the FB inserts, thus negatively affecting the correlation between damage score and deviation. This would suggest that these correlations might be strengthened by examining the correlations across many, much smaller regions of the articular surfaces. We are currently investigating the use of image stitching techniques to perform such correlations on a pixel-by-pixel basis between digital photographs of the articular surfaces (rather than subjective scores) and deviation measurements. As to our second research question, the deviation patterns did vary between the RP and FB inserts. The symmetric distribution of deviations between the medial and lateral plateaus of the RP design reflects the higher conformity of the tibiofemoral articulation coupled with the rotation allowed at the insert–tray interface. This motion allows both femoral condyles to be in equal contact with the polyethylene insert. The lack of anterior post “bowtie” damage in the RP inserts is consistent with our previous study of a larger cohort of RP versus FB inserts [2,12], and demonstrates that less impingement with the intercondylar box of the femur occurs when the insert is allowed to rotate. The more axisymmetric deviation patterns found in the FB inserts likely reflect the lower tibiofemoral conformity and the fact that the insert is locked in the tray. These features allow the femur to slide across the polyethylene insert. The higher deviations on the medial plateau are consistent with higher loads being transferred through the
medial plateau, as has been shown with loads measured from instrumented TKAs [15]. Additionally the “bowtie” pattern present on many of the FB tibial posts indicate that unlike the RP, the femoral component impinges with the post, affecting the translation and rotation of the femoral relative to the tibial component. Our study has several limitations. First, it is a retrieval study, so our evaluation was limited to failed implants, many of which were revised for reasons (such as stiffness, osteolysis, and loosening) associated with early failure and abnormal wear. Second our cohort is small, which may have limited our ability to determine differences. Third, the original dimensions of the inserts were unknown, so we relied on comparison to pristine inserts of the same design and manufacturer. Nonetheless, this precluded us from determining the absolute volumetric wear even assuming manufacturing tolerances [9]. Finally, the 3D laser scanning system measures the deformation in the retrieved components; however; it cannot distinguish deviations due from plastic deformation from those caused by material loss through wear. In summary, the commercially available table top scanner could measure differences in deviations between two similar, but different implant designs. Most importantly, with the scanner we could obtain subjective data over a larger area, allowing us to determine differences in in vivo performance between RP and FB implants. This coupled with the availability, ease of use, speed, and low cost makes it a useful tool for modern retrieval analysis. Acknowledgment We thank Marcella Elpers for her assistance in implant retrieval. We also thank Seth Jerabek, MD, Stephanie Tow, and Rose Fu for assistance with implant grading. References 1. Hood RW, Wright TM, Burstein AH. Retrieval analysis of total knee prostheses: a method and its application to 48 total condylar prostheses. J Biomed Mater Res 1983;17:829. 2. Kelly NH, Fu RH, Wright TM, et al. Wear damage in mobile-bearing TKA is as severe as that in fixed-bearing TKA. Clin Orthop Relat Res 2011;469:123.
156
K.E. Stoner et al. / The Journal of Arthroplasty 28 Suppl. 1 (2013) 152–156
3. Dolan MM, Kelly NH, Nguyen JT, et al. Implant design influences tibial post wear damage in posterior-stabilized Knees. Clin Orthop 2011;469:160. 4. Tsao A, Mintz L, McRae C, et al. Failure of the porous-coated anatomic prosthesis in total knee arthroplasty due to severe polyethylene wear. J Bone Joint Surg 1993;75:19. 5. Medel FJ, Kurtz SM, Parvizi J, et al. In vivo oxidation contributes to delamination but not pitting in polyethylene components for total knee arthroplasty. J Arthroplasty 2011;26:802. 6. Rawlinson JJ, Furman BD, Li S, et al. Retrieval, experimental, and computational assessment of the performance of total knee replacements. J Orthop Res 2006; 24:1384. 7. Engh GA, Zimmerman RL, Parks NL, et al. Analysis of wear in retrieved mobile and fixed bearing knee inserts. J Arthroplasty 2009;24:28. 8. Berry DJ, Currier JH, Mayor MB, et al. Knee wear measured in retrievals: a polished tray reduces insert wear. Clin Orthop Relat Res 2012;470:1860. 9. Teeter MG, Douglas D, Naudie R, et al. In vitro quantification of wear in tibial inserts using microcomputed tomography. Clin Orthop Relat Res 2011;469:107.
10. Teeter MG, Naudie DDR, Milner JS, et al. Determination of reference geometry for polyethylene tibial insert wear analysis. J Arthroplasty 2011;26:497. 11. Engh CA, Zimmer RL, Hopper RH, et al. Can microcomputed tomography measure retrieved polyethylene wear? Comparing fixed-bearing and rotating-platform knees. Clin Orthop Relat Res 2013;471:86. 12. Stoner K, Jerabek SA, Tow S, et al. Rotating-platform has no surface damage advantage over fixed-bearing TKA. Clin Orthop Relat Res 2013;471:76. 13. Harman MK, DesJardins J, Benson L, et al. Comparison of polyethylene tibial insert damage from in vivo function and in vitro wear simulation. J Orthop Res Off Publ Orthop Res Soc 2009;27:540. 14. Estupiñán JA, Bartel DL, Wright TM. Residual stresses in ultra-high molecular weight polyethylene loaded cyclically by a rigid moving indenter in nonconforming geometries. J Orthop Res 1998;16:80. 15. Mündermann A, Dyrby CO, D’Lima DD, et al. In vivo knee loading characteristics during activities of daily living as measured by an instrumented total knee replacement. J Orthop Res 2008;26:1167.