A multimodal assessment of cementless tibial baseplate fixation using radiography, radiostereometric analysis, and magnetic resonance imaging

Fixation in cementless total knee arthroplasty is provided by osseous integration. Radiography, radiostereometric analysis (RSA), and magnetic resonance imaging (MRI) were used simultaneously to investigate fixation. Relationships between RSA‐measured implant micromotions and MRI‐evaluated osseous integration at the component–bone interface were assessed in 10 patients up to 6 months postoperation. Supine MRI (using multispectral imaging sequences) and RSA exams were performed to evaluate osseous integration and measure longitudinal migration, respectively. Inducible displacement was measured from standing RSA exams. Radiolucent lines were detected on conventional radiographs. Of 10 patients, 6 had fibrous membranes detected on MRI. No fluid or osteolytic interfaces were found, and no components were scored loose. Of 10 patients, 6 had radiolucent lines detected. Average maximum total point motion (MTPM) for longitudinal migration at 6 months was 0.816 mm (range 0.344–1.462 mm). Average MTPM for inducible displacement at 6 months was 1.083 mm (range 0.553–1.780 mm). Fictive points located in fibrous‐classified baseplate quadrants had greater longitudinal migration than fictive points located in baseplate quadrants with normal interfaces at 2 weeks (p = 0.031), 6 weeks (p = 0.046), and 3 months (p = 0.047), and greater inducible displacements at 3 months (p = 0.011) and 6 months (p = 0.045). Greater early micromotion may be associated with the presence of fibrous membranes at the component–bone interface. Clinical significance: This multimodal imaging study contributes knowledge of the fixation of modern cementless TKA, supporting the notion that osseous integration is important for optimal implant fixation.


| INTRODUCTION
[3] Osseous integration can provide long-term fixation without the need for bone cement and its associated disadvantages. 3,4Early reports are encouraging and support the use of cementless devices; however, the role of design features such as pegs, stems, porosity, and coefficient of friction on initial implant stability and how early bone ingrowth develops in cementless TKA requires additional study. 5,68][9][10][11] Radiostereometric analysis (RSA) is an established tool for evaluating in vivo fixation, and is capable of tracking implant micromotion through as early as the first 6 months following arthroplasty to assess early implant stability and predict future loosening risk. 124][15][16][17][18] However, RSA does not assess the osseous integration that provides the fixation.Radiographic techniques can provide information on the component-bone interface through the presence of periprosthetic radiolucencies, but there is little information on the diagnostic performance of radiography to assess implant loosening. 191][22][23] Improved techniques for reducing the encoding distortions induced by the metallic components have made it possible to assess the component-bone interface, allowing radiologists to classify an implant as loose based on the quality and quantity of osseous integration.Concurrent RSA and MRI exams could take advantage of the strengths of both techniques and provide complementary information on the fixation status of cementless devices.Therefore, the goal of this prospective cohort study was to investigate RSA-measured cementless tibial baseplate micromotion and the relationship to MRI-evaluated osseous integration at the corresponding component-bone interfaces.It was hypothesized that tibial baseplates with good osseous integration would have less micromotion than baseplates with poorer osseous integration.

| METHODS
This was a prospective cohort study (levels of evidence-Level II) that sequences to reduce susceptibility artifact. 24,25Pulse sequences included coronal T2-weighted fast spin echo sequences, and axial and sagittal inversion recovery and proton density-weighted threedimensional (3D) MAVRIC-SL sequences.MRI scans were scored by an experienced musculoskeletal radiologist (H.G.P.).Component-bone interfaces were separately evaluated for four quadrants of the tibial baseplate tray (anterolateral, anteromedial, posterolateral, posteromedial), similar to the method used by Mosich et al. 23 Quadrants were classified as normal or containing a fibrous membrane, fluid interface, or osteolysis, and the osseous integration was scored as a percentage of the quadrant occupied by a normal interface (<33%, 33%-66%, and >66%).The presence of periprosthetic bone marrow edema patterns were noted, and a final verdict of loose was determined if there was almost no (<5%) normal interface along the component-bone interface.
The inter-reader reproducibility of MRI has previously been shown to be substantial-to-excellent for interface characterization (Gwet agreement coefficient, 0.75-0.89)and for degree of osseous integration (Gwet agreement coefficient, 0.67-0.96).There is near-perfect reproducibility for characterizing each component as loose at MRI (Gwet agreement coefficient, 0.90, 0.95, and 0.93 for patellar, femoral, and tibial components, respectively). 22andardized supine RSA exams occurred within 2 days postoperation (baseline), and at 2 weeks, 6 weeks, 3 months, and 6 months postoperation using a biplanar calibration cage (RSA Biomedical). 26A second supine exam was performed at the 6-week visit to assess bias and repeatability in the methods described by Niesen and Hull. 27Standing, weightbearing RSA exams were acquired at all study visits excluding the baseline appointment using | 101 displacements were measured from micromotion occurring between a standing and supine exam from the same study visit.Migrations and displacements were reported as maximum total point motions (MTPM), and as lateral-medial, superior-inferior, anterior-posterior, and resultant 3D translations at fictive points located around the tibial baseplate (Figure 1). 13Lateral, superior, and anterior directions were defined as positive translations.
Standing hip-knee-ankle images were acquired preoperatively and at 6 weeks postoperation and were used to measure coronal alignment in a method similar to Cooke et al. 28

| RESULTS
A total of 10 participants were included and analyzed.Demographic information, as well as preoperative and postoperative coronal alignment is displayed in Table 1.
Fibrous interfaces (example shown in Figure 2) were identified on MRI at the component-bone interface for at least one study visit in 6 of 10 patients (Figure 3).Two patients had unreadable scans at 6 weeks postoperation due to motion and enhanced susceptibility.The bias and repeatability for artifactual migration were 0.375 and 0.195 mm, respectively, for MTPM and were 0.220 and 0.137 mm, respectively, for 3D fictive point translations.For differences in apparent MTPM, the bias and repeatability were 0.008 and 0.198 mm, respectively, whereas for apparent differences in 3D fictive point translations, bias and repeatability were 0.019 and 0.118 mm, respectively.The patterns of anterior-posterior, superior-inferior, and lateral-medial migrations of fictive points are depicted in Figure S1.
Patterns of anterior-posterior, superior-inferior, and lateral-medial inducible displacements of fictive points are displayed in Figure S2.
Patterns for both migration and displacement indicate axial rotation, either external or internal, is the predominate mode of micromotion.
A subgroup analysis of longitudinal migration was performed using the component-bone interface classifications to split patients into those who had a fibrous membrane within the first 6 weeks postoperation and those who had all normal interfaces in the first 6 weeks (Figure 4).Average MTPM progressed from 0.445 mm at 2 weeks to 0.611 mm at 6 weeks, 0.592 mm at 3 months, and 0.583 mm at 6 months for the normal group, and from 0.645 mm at 2 weeks to 0.913 mm at 6 weeks, 1.005 mm at 3 months, and 1.049 mm at 6 months for the fibrous group.Two-way ANOVA found no significant difference (p = 0.102) for migration MTPM of the tibial baseplate between fibrous and normal classified patients.
However, two-way ANOVA did find a significant difference (p < 0.0001) in migrations between fictive points in fibrousclassified quadrants compared to fictive points found in quadrants classified as normal, excluding the stem tip and lateral or medial points located between a normal and fibrous quadrant (Figure 5).[16][17][18]29,30 Radiographic assessment for radiolucent lines was also comparable to other cementless TKA. 19bgroup analyses found significantly greater longitudinal migration and inducible displacement at fictive points located in quadrants containing fibrous membranes to fictive points located in normally classified quadrants, supporting the hypothesis that baseplates with good osseous integration have less micromotion.
Normally, RSA studies follow patients up to 2 years postoperation, where established thresholds for 6-month migration and continuous migrations from 6 months to 1 year and 1-2 years can be used to predict future risk of loosening. 12,31The current study followed patients up to 6 months since the focus was on the relationship between early stabilization of baseplates and the osseous integration that provides the biological fixation, which develops by 6 months postoperation. 32,334][15][16] Inducible displacements measured in this study are also comparable to displacements in implants with good fixation. 29,30It is therefore reasonable to expect components in this study to be similarly well-fixed.
The findings from MRI support the RSA measurements indicating that strong fixation is achieved.A study diagnosing TKA component loosening using MRI by Endo et al. revealed that loosening is associated with fluid interfaces, osteolysis, complete absence of normal (trabecular bone signal flush with the metal or cement) interfaces, and poor (<33%) osseous integration. 22None of the 10 tibial baseplates in this study met any of those criteria, as there were only normal or fibrous interfaces present and no patient had more than two quadrants classified as fibrous at any given time point.
At 6 months 4 of 40 tibial quadrants were fibrous, a ratio similar to another MRI study of a posterior-stabilized version of the same cementless implant. 23The presence of fibrous membranes results in the stimulation of osteoclasts that can lead to bone resorption and osteolysis, and indicates limited osseous integration has been achieved. 21,22,34,35Again, no signs of osteolysis were present in this study, and most patients had their fibrous membranes resolved by 6 months.However, further follow-up would be needed to confirm an absence of osteolysis development in the future.The exact amount of osseous integration that is required for successful fixation of cementless tibial baseplates is unknown; however, a study of retrieved baseplates by Purcell et al. found a higher percentages of bone ingrowth in modern cementless designs (64%) than has been demonstrated in earlier generation designs (<30%). 2,33,36Almost every quadrant of all baseplates in the current study had >66% osseous integration measured on MRI, suggesting similar rates of bone ingrowth and strong fixation can be expected.
A similar incidence of small (1 mm) radiolucent lines were observed around baseplates as in a previous study of cementless TKA, which demonstrated that while common, these radiolucencies are often nonprogressive, can resolve, and are present even in wellfunctioning knees. 19Four of the six patients with radiolucent lines in the current study had fibrous membranes detected, with the other two having all normal interfaces on MRI.Endo et al. observed similar performance between radiography and MRI for diagnosing loosening in tibial components. 22The radiographic and MRI findings are consistent here as well, with both suggesting that minor interference at the component-bone interface can still be observed in well-fixed cementless implants.
It is thought that early micromotion can prevent osseous integration of cementless tibial components and lead to the formation of fibrous membranes. 2,34,36The average migration measured at the fictive points located in regions with fibrous membranes within the first 6 weeks postoperation was noticeably greater than the average for fictive points located in all normal component-bone interfaces in the first 6 weeks.The discrepancy in migration occurred in the first 6 weeks, the first phase of the biphasic migration pattern that relies on the initial press-fit and auxiliary fixation structures such as pegs.Most baseplates, regardless of early interface classification, reached a plateau in migration after 3 months suggesting secondary fixation via bone ingrowth is eventually achieved.However, there is an increased risk associated with greater early migration, highlighting the importance of achieving good early stability in cementless devices. 12,37ny modern cementless designs implement pegs that are intended to provide stability while also having the capacity for bone ingrowth. 2As the predominate pattern of micromotion observed for this particular implant is axial rotation, the pegs may be crucial for early stabilization. 38Purcell et al. found less osseous integration in the medial pegs than in the lateral pegs, explained by mechanical and biological changes in the medial bone of varus osteoarthritic knees. 2,39,40Most knees in the current study were varus preoperatively, and while component-bone interfaces for the pegs were not assessed, fibrous membranes were more common on the medial surfaces.
This study was not without limitations.First, the sample size was small, as this was a prospective cohort study with the goal of finding relationships between MRI-evaluated osseous integration and RSAmeasured micromotion that placed a high burden on recruited subjects.Another limitation is that the available MR scanner had a field strength of 3.0 T. Magnetic susceptibility artifact, which causes in-plane and through-plane image distortion and the presence of pixel pile-up and signal voids, is directly proportional to magnetic field strength, making the component-bone interfaces of metal knee replacements more difficult to interpret.In this instance, a 1.5 T scanner would be optimal; however, the scans were read by a musculoskeletal radiologist with experience reading images acquired at both 1.5 and 3.0 T. Motion artifact was also an issue affecting image quality.Patients were scanned early in their recovery and had difficulty remaining motionless for the individual imaging series which had scan times of approximately 8-12 min.However, only two scans were unreadable.Furthermore, it is possible that less than 2 days is too short of a time for fibrous membranes to form; however, in cases where component-bone interfaces were classified as fibrous at the first follow-up there was still a clear lack of osseous signal immediately adjacent to the opposing implant surface.In addition, two surgeons performed the TKA procedures, introducing surgeon variability, although both use the cementless device investigated here regularly in their standard practice.
This prospective evaluation of cementless tibial components using RSA and MRI at paired time points observed the progressive fixation of baseplates.Longitudinal migration and inducible displacement were comparable to previous studies and component-bone interfaces were mostly normal with good osseous integration, and there were no components scored as loose, suggesting baseplates were well-fixed.Tibial baseplates that had a fibrous membrane detected within the first 6 weeks postoperation had increased longitudinal migration up to 6 months, mainly due to high initial migration in the first 6 weeks.This study contributes knowledge of the performance of modern cementless total knee implants, supporting the notion that osseous integration is important for optimal implant fixation.
followed patients up to 6 months postoperation.Ethics approval was obtained from an Institutional Review Board (Lawson Approval Number R-20-143).Participants were prospectively recruited at University Hospital, London Health Sciences Centre between October 2020 and December 2021.Inclusion criteria were patients over the age of 18 years with knee osteoarthritis scheduled for a unilateral primary TKA.Patients were excluded if they received a diagnosis of inflammatory arthritis, had prior knee surgery, were or planned on becoming pregnant, had poor bone quality (determined preoperatively or intraoperatively by surgeon), a cognitive impairment, an inability to understand English, a history of alcoholism, a contraindication for 3.0 T MRI, or were over the age of 75 years.Two fellowship-trained arthroplasty surgeons (B.A.L. and J.L.H.) performed operations using a standard midline incision and a medial parapatellar arthrotomy.A femur first technique was used, as were extramedullary tibial guides.A cementless, single radius, cruciateretaining TKA implant design was used in all cases (Triathlon Tritanium; Stryker), consisting of a cobalt-chromium femoral component, an additively manufactured titanium tibial baseplate, and a condylar stabilizing polyethylene insert.Patellar resurfacing was performed with a metal-backed cementless patellar component.The target was neutral mechanical alignment.Up to 8 tantalum markers were inserted into the proximal tibia intraoperatively to enable micromotion analysis using RSA.The postoperative care protocol was identical for all patients.A 3.0 T clinical scanner (General Electric Healthcare) was used for MRI evaluation within 2 days postoperation, at 2 weeks, 6 weeks, 3 months, and 6 months postoperation.An 8-channel knee coil or 32-channel cardiac coil was used depending on knee size.The imaging protocol utilized multi-acquisition with variable resonance image combination selective (MAVRIC-SL; General Electric Healthcare) a uniplanar calibration cage (RSA Biomedical).Model-based RSA software (RSAcore) was used to analyze the RSA images.Longitudinal migrations were measured from implant micromotion occurring between a supine follow-up exam and the baseline exam.Inducible BROBERG ET AL.
Standard standing anteroposterior (AP) and lateral radiographs were acquired at 6 weeks postoperation and were used to detect radiolucent lines in a method similar to Costales et al., with the medial tibial tray, lateral tibial tray, and regions surrounding the stem on the AP radiograph and the anterior tibial tray, posterior tibial tray, and stem tip on the lateral radiograph evaluated for radiolucent lines of 1, 2, or >2 mm. 19Statistical analyses were performed using Prism 9 (GraphPad Software).Normality was assessed with a D'Agostino and Pearson omnibus normality test.Two-way analysis of variance (ANOVA) was used to assess longitudinal migration between normal and fibrous classified patients and fictive points.Inducible displacements between normal and fibrous classified fictive points were compared using Mann-Whitney tests.Preoperative and postoperative outcome scores and coronal alignment were compared using a paired t-test or Wilcoxon matched-pairs signed rank test depending on normality.

Four of 40
quadrants were fibrous at <2 days, 6 of 40 quadrants were fibrous at 2 weeks, 1 of 32 quadrants were fibrous at 6 weeks, 2 of 40 quadrants were fibrous at 3 months, and 4 of 40 quadrants were fibrous at 6 months postoperation.Five of the six patients with quadrants classified as fibrous had fibrous membranes detected on F I G U R E 1 Inferior view of a three-dimensional model of the right tibial component depicting the location of the fictive points (purple spheres) labeled L: lateral, AL: anterolateral, ALP: anterolateral peg, AM: anteromedial, AMP: anteromedial peg, M: medial, PL: posterolateral, PLP: posterolateral peg, PM: posteromedial, PMP: posteromedial peg, and ST: stem tip.T A B L E 1 Patient demographics and coronal alignment.

F I G U R E 3
Component-bone interface classifications for all 10 tibial baseplates across all study visits.Two scans were unreadable (Patient 8, 6 weeks; Patient 9, 6 weeks).All baseplates have been normalized to right knees.Baseplates were divided into four quadrants (anterolateral, anteromedial, posterolateral, and posteromedial).White regions indicate normal interfaces and purple regions indicate a fibrous membrane classification.Asterisks indicate patients with radiolucent lines present on anteroposterior (AP) and/or lateral radiographs at 6 weeks.A, anterior; L, lateral; M, medial; P, posterior.
Šidák's multiple comparisons test found differences at 2 weeks (p = 0.031), 6 weeks (p = 0.046), and 3 months (p = 0.047).A subgroup analysis of inducible displacement was performed using the component-bone interface classifications to split fictive points into those located in a quadrant with a fibrous membrane and those in normal interfaces (excluding the stem tip and lateral or medial points between a normal and fibrous quadrant) at the time of the inducible displacement exam (Figure6).Average MTPM for inducible displacement was 0.795 mm at 2 weeks, 1.394 mm at 6 weeks, 0.873 mm at 3 months, and 0.972 mm at 6 months for normal classified patients, and was 0.923 mm at 2 weeks, 0.875 mm at 6 weeks, 1.378 mm at 3 months, and 1.524 mm at 6 months for the fibrous classified patients.Significantly greater 3D translations between fictive points in fibrous-classified quadrants compared to fictive points found in quadrants classified as normal were found at 3 months (p = 0.011) and 6 months (p = 0.045).F I G U R E 4 Plot of longitudinal migration maximum total point motions (MTPM) from baseline to 6 months postoperation.Purple indicates patients with a fibrous membrane detected within 6 weeks of surgery and gray denotes patients that had all normal interfaces in the first 6 weeks.Solid lines depict individual patient migrations, dashed lines indicate the average migrations, and shaded regions highlight the range.The black arrow identifies the patient with all normal interfaces within the first 6 weeks who later developed fibrous membranes at 3 and 6 months postoperation.F I G U R E 5 Fictive point three-dimensional (3D) translations for longitudinal migration at 2 weeks, 6 weeks, 3 months, and 6 months postoperation.Purple indicates fictive points in quadrants with a fibrous membrane detected within 6 weeks of surgery and gray denotes fictive points in quadrants that had all normal interfaces in the first 6 weeks.AL, anterolateral; ALP, anterolateral peg; AM, anteromedial; AMP, anteromedial peg; L, lateral; M, medial; PL, posterolateral; PLP, posterolateral peg; PM, posteromedial; PMP, posteromedial peg; ST, stem tip.This is the first study to implement RSA and MRI concurrently to examine implant fixation in the same cohort of patients.The MRI results revealed most patients had excellent osseous integration and no baseplates were scored as loose.RSA measurements indicated that fixation was occurring between the implant and bone, with longitudinal migration patterns often stabilizing by 3 months postoperation and inducible displacement values comparable to those of

F I G U R E 6
Maximum total point motion (MTPM) and fictive point three-dimensional (3D) translations for inducible displacement at 2 weeks, 6 weeks, 3 months, and 6 months postoperation.Purple indicates MTPM of patients with a fibrous membrane detected at the time of the displacement exam, or fictive points in quadrants that contained a fibrous membrane at the time of the displacement exam.Gray denotes MTPM of normal classified patients, or fictive point translation in quadrants that were classified as normal.AL, anterolateral; ALP, anterolateral peg; AM, anteromedial; AMP, anteromedial peg; L, lateral; M, medial; PL, posterolateral; PLP, posterolateral peg; PM, posteromedial; PMP, posteromedial peg; ST, stem tip.BROBERG ET AL. | 105