Do Metaphyseal Cones and Stems Provide Any Biomechanical Advantage for Moderate Contained Tibial Defects in Revision TKA? A Finite-Element Analysis Based on a Cadaver Model

Abstract

Background

Satisfactory management of bone defects is important to achieve an adequate reconstruction in revision TKA. Metaphyseal cones to address such defects in the proximal tibia are increasingly being used; however, the biomechanical superiority of cones over traditional techniques like fully cementing the implant into the defect has not yet been demonstrated. Moreover, although long stems are often used to bypass the defects, the biomechanical efficacy of long stems compared with short, cemented stems when combined with metaphyseal cones remains unclear.

Questions/purposes

We developed and validated finite-element models of nine cadaveric specimens to determine: (1) whether using cones for addressing moderate metaphyseal tibial defects in revision TKA reduces the risk of implant-cement debonding compared with cementing the implant alone, and (2) when using metaphyseal cones, whether long, uncemented stems (or diaphyseal-engaging stems) reduce the risk of implant-cement debonding and the cone-bone micromotions compared with short, cemented stems.

Methods

We divided nine cadaveric specimens (six male, three female, aged 57 to 73 years, BMI 24 to 47 kg/m²) with standardized tibial metaphyseal defects into three study groups: no cone with short (50-mm) cemented stem, in which the defect was filled with cement; cone with short (50-mm) cemented stem, in which a metaphyseal cone was implanted before cementing the implant; and cone with long, diaphyseal-engaging stem, which received a metaphyseal cone and the largest 150-mm stem that could fit the diaphyseal canal. The specimens were implanted and mechanically tested. Then, we developed and validated finite-element models to investigate the interaction between the implant and the bone during the demanding activity of stair ascent. We quantified the risk of implant debonding from the cement mantle by comparing the axial and shear stress at the cement-implant interface against an experimentally derived interface failure index criterion that has been previously used to quantify the risk of cement debonding. We considered the risk of debonding to be minimal when the failure index was below 10% of the strength of the interface (or failure index < 0.1). We also quantified the micromotion between the cone and the bone, as a guide to the likelihood of fixation by bone ingrowth. To this end, we assumed bone ingrowth for micromotion values below the most restrictive reported threshold for bone ingrowth, 20 µm.

Results

When using a short, 50-mm cemented stem and cement alone to fill the defect, 77% to 86% of the cement-implant interface had minimal risk of debonding (failure index < 0.1). When using a short, 50-mm cemented stem with a cone, 87% to 93% of the cement-implant interface had minimal debonding risk. When combining a cone with a long (150-mm) uncemented stem, 92% to 94% of the cement-implant interface had minimal debonding risk. The differences in cone-bone micromotion between short, cemented stems and long, uncemented stems were minimal and, for both configurations, most cones had micromotions below the most restrictive 20-µm threshold for ingrowth. However, the maximum micromotion between the cone and the bone was in general smaller when using a long, uncemented stem (13-23 µm) than when using a short, cemented stem (11-31 µm).

Conclusion

Although the risk of debonding was low in all cases, metaphyseal cones help reduce the biomechanical burden on the implant-cement interface of short-stemmed implants in high-demand activities such as stair ascent. When using cones in revision TKA, long, diaphyseal-engaging stems did not provide a clear biomechanical advantage over short stems. Future studies should explore additional loading conditions, quantify the interspecimen variability, consider more critical defects, and evaluate the behavior of the reconstructive techniques under repetitive loads.

Clinical Relevance

Cones and stems are routinely used to address tibial defects in revision TKA. Despite our finding that metaphyseal cones may help reduce the risk of implant-cement debonding and allow using shorter stems with comparable biomechanical behavior to longer stems, either cones or cement alone can provide comparable results in contained metaphyseal defects. However, longer term clinical studies are needed to compare these techniques over time.

Introduction

Moderate bone defects involving loss of metaphyseal bone, Type II-b in the Anderson Orthopaedic Research Institute classification system, are often present during revision TKA. Long-term fixation of components during revision TKA requires adequate management of such metaphyseal bone defects. To this end, several techniques have been proposed, including impaction grafting, structural allograft, filling the defect with cement, and, more recently, filling the defect with metaphyseal cones into which the implant is cemented [20]. Although metaphyseal cones are increasingly popular, one recent study found that metaphyseal cones had no superior survivorship or patient-reported outcomes over a hybrid cementing technique where the defect is filled with cement [5]. Thus, despite the widespread clinical use of cones to address metaphyseal defects, it is unclear whether cones provide any biomechanical advantage over traditional techniques. In addition, stems are commonly used as part of the reconstruction to bypass the defects and unload the most compromised areas of the bone [3]; however, one recent study reported better outcomes for short, cemented stems compared with long, uncemented stems when combined with metaphyseal cones [7]. Therefore, the biomechanical efficacy of long stems when combined with metaphyseal cones remains unclear. Two main concerns with these fixation techniques are the implant debonding from the surrounding cement mantle [1, 10, 18] and achieving bone ingrowth into the metaphyseal cone. From a biomechanical standpoint, cement debonding is related to high stresses at the interface between the implant and cement [11], while bone ingrowth requires minimal relative motion between the surface of the cone and the surrounding bone. The threshold for bone ingrowth was determined in canine models to be in the range of 20 to 150 mm of micromotion [8, 9].

To assess the biomechanical behavior of different reconstruction techniques in revision TKA, researchers have used both in vitro mechanical testing and computational finite-element models [2, 13, 17, 28, 29] with contradictory results. Prior studies considering synthetic bone reported reduced micromotion with respect to the bone for sleeves [13] and cones [28] when using long stems. In this way, despite their fundamental design differences, cones and sleeves showed similar biomechanical behavior in these studies. However, one study with a CT-derived finite-element model reported no reduction of micromotion with long stems [2]. Although these studies provided insights into the biomechanics of implants for revision TKA, their conclusions are limited by their use of a single, often synthetic bone that does not capture the variation in geometry and distribution of material properties of actual patients undergoing revision TKA. Furthermore, except for Awadalla et al. [2], their use of a few simplified load cases makes it unlikely that they captured the worst-case scenario for the interaction between the implant components, the cement, and the bone, which often involves submaximal forces and large moments [15, 23]. Moreover, the potential biomechanical advantage provided by metaphyseal cone augmentation over traditional techniques, like filling the defect with cement, has not been established.

In this study, we simulated the demanding activity of stair ascent on finite-element models developed from the CT scans of nine cadaveric specimens implanted with revision TKA components to determine: (1) whether using cones for addressing moderate metaphyseal tibial defects in revision TKA reduces the risk of implant-cement debonding compared with cementing the implant alone, and (2) when using metaphyseal cones, whether long, uncemented stems (or diaphyseal-engaging stems) reduce the risk of implant-cement debonding and the cone-bone micromotions compared with short cemented stems. To maximize the contribution of the cone in the first research question, we considered short, cemented stems.

Materials and Methods

Study Overview

The study was designed in two parts. We first implanted and mechanically tested three sets of three matched cadaveric tibiae under axial loading. Three configurations addressed the same metaphyseal defect: a defect filled with cement alone (no cone) but with the addition of a short (50-mm), fully cemented stem; a defect filled with a cone with the addition again of a short, fully cemented stem; and a defect filled with a cone with the addition of a long, diaphyseal-engaging stem with hybrid cementation technique. Then, we created individual finite-element models of the nine cadaveric specimens that reproduced the experimental implantation. The finite-element models were validated against experimental mechanical testing of the same implanted cadaveric tibiae and used to evaluate the risk of cement debonding from the implant and to determine the interfacial micromotion at the cone-bone interface intended for ingrowth during simulated stair ascent (Fig. 1). The risk of debonding was quantified with the interfacial failure index, calculated from the interfacial normal and shear stresses at the cement-implant interface [30]. A failure index above 1 indicated debonding, while we assumed a failure index below 0.1 (a safety factor of 10) represented minimal risk of debonding. For micromotion, we considered the most restrictive proposed threshold of micromotion that would result in bone ingrowth, 20 µm [8].

Fig. 1.

Overall workflow of the study. Cadaveric specimens CT scans were taken pre- and postimplantation and the corresponding bone geometries registered to generate finite-element models. The specimens were tested experimentally, and the surface strains were compared against those predicted with the FE models. The validated FE models were used to simulate the activity of stair ascent and compute the risk of debonding of the implant from the cement mantle and the micromotion between the cone and the bone; FE = finite element.

Specimen Preparation

We procured nine fresh-frozen cadaveric tibia specimens of representative ages for TKA (57 to 73 years old), without gross bone defects or signs of prior hardware. The specimens were divided into three sets of three specimens. Each set consisted of both tibiae from the same donor and a third specimen that was matched based on height, weight, and sex (Table 1).

Table 1.

Demographics of the cadaveric human specimens and selected implant sizes for the three study groups and for the three specimen sets

Specimen set	1			2			3
Group	No cone	Cone, short stem	Cone, long stem	No cone	Cone, short stem	Cone, long stem	No cone	Cone, short stem	Cone, long stem
Age in years	71	71	72	73	73	57	62	62	65
Sex	Male	Male	Male	Female	Female	Female	Male	Male	Male
Side	Right	Left	Right	Left	Right	Left	Right	Left	Right
Weight in kg	145	145	145	95	95	91	78	78	89
Height in cm	175	175	180	168	168	168	178	178	175
Cone sizea	-	E	E	-	C	C	-	E	E
Baseplate sizeb	4	4	6	2	2	3	4	4	4
Stem length in mm	50	50	150	50	50	150	50	50	150
Stem diameter in mm	12	12	18	12	12	16	12	12	14

^aMeasured dimensions of Stryker Triathlon cones: Size C: 36.5 mm (proximal diameter), 25 mm (distal diameter), 29 mm (height); Size E: 40.5 mm (proximal diameter), 25 mm (distal diameter), 39 mm (height).

^bMeasured dimensions of Stryker Triathlon baseplates: Size 2: 42 mm (AP), 64 mm (ML); Size 3: 44 mm (AP), 67 mm (ML); Size 4: 46 mm (AP), 70 mm (ML); Size 6: 52 mm (AP), 77 mm (ML).

Two fellowship-trained orthopaedic surgeons (NS, KGM) prepared and implanted the specimens with commercially available TKA tibial components (Triathlon, Stryker) under intramedullary guidance and using standard instruments, according to the surgical technique provided by the manufacturer. The proximal tibial resection was standardized for all specimens to be 10 mm below the center of the lateral plateau to represent the situation during revision TKA. The resection was perpendicular to the mechanical axis of the tibia in the coronal plane and with 3° of posterior slope in the sagittal plane.

After resection, the surgeons created a metaphyseal defect by reaming to the largest cone that could fit the metaphysis. All three specimens within a set had the same size symmetric defect created with the manufacturer’s specially designed reamers. One tibia from each set was assigned to the no cone with short, cemented stem group and was prepared to receive a short, 50-mm-long x 12-mm-diameter stem that was fully cemented into the bone such that the cement filled the defect (Fig. 2A). This is the smallest stem offered by the system and was chosen to maximize the effect of the cone. Another tibia from each set was assigned to the cone with short, cemented stem group and a metaphyseal cone was implanted before cementing a 50-mm-long x 12-mm-diameter stem into the bone-cone construct (Fig. 2B), such that the space between the stemmed implant and the cone was filled with cement. The third specimen from each group was assigned to the cone with long diaphyseal-engaging stem group and was prepared by reaming the canal until the reamer began to chatter to determine the largest diameter 150-mm-long stem that could fit and fill the diaphyseal canal (Fig. 2C). For this group, a cone was implanted using a hybrid technique, where the cement only covered the underside of the baseplate and the implant portion within and proximal to the cone (that is, most proximal 35 mm).

Fig. 2.

A-C Coronal cross-sectional view of the finite-element models for the three configurations studied: (A) no cone with short, cemented stem; (B) cone with a short, cemented stem; and (C) cone with a long, diaphyseal-engaging stem. The visible portion of the cement mantle is indicated. For the configurations with a cone (B, C), cement is also present within the metaphyseal cone. The load application fixture and the potting are also shown. For the stair ascent cycle, the potting was replaced by a fixed boundary constraint, while the load application fixture was replaced by a coupling constraint to apply the forces and moments.

We used symmetric Triathlon Tritanium cones (Stryker). These feature a layer of three-dimensional (3D)-printed highly porous titanium on the outer surface to allow for bone ingrowth.

After preparation and before implantation of the final components, the specimens were CT-scanned alongside bone density reference phantoms (140 mA, 140 kVp, and 0.6-mm slice thickness; Biograph-64 CT scanner, Siemens Healthcare). The scans were used to obtain the geometry of the bones with the created metaphyseal defects and the distribution of bone material properties for the finite-element models.

The specimens were then implanted according to the group they were assigned, using the appropriate baseplate size to maximize coverage and limit overhang (Table 1). Because we tried to create the largest defect possible, the final implant sizes were smaller than those recommended by the manufacturer based on the cone size. However, all chosen implant-cone combinations were still acceptable according to manufacturer’s guidelines. The cone and the baseplate were manually impacted, according to the manufacturer’s guidelines. After implantation, three pairs of retroreflective markers were attached to the bone and the implant along the anterior aspect of the bone-implant interface (Fig. 3A). A second CT scan of each specimen after implantation was used to determine the position of the implanted components within the preimplantation bones. The retroreflective markers, visible in the second CT scan, were measured at the beginning of the testing using a motion capture system (Cortex, Motion Analysis Corp) and were used in the finite-element models to replicate the position of the specimen with respect to the fixture and actuator during experimental testing.

Fig. 3.

A-B (A) Experimental test setup for model validation; (B) finite-element model for validation.

Mechanical Testing

The specimens were resected approximately 285 mm distal to the baseplate-bone interface. The most distal 70 mm of each specimen were potted in epoxy (Bondo, 3M) and mounted on a servo-hydraulic testing machine (Model 825, MTS) (Fig. 3A). Additional retroreflective markers were placed on the fixture of the machine to identify the location of loading for the finite-element models. To measure the bone surface strains at critical locations that were then used for the validation of the finite-element models, we instrumented all nine specimens with two strain gauge stacked rosettes (C2A-06-031WW-120, Micro Measurements, Vishay Precision Group): one on the posterior-medial side, approximately aligned with the loading actuator, and another on the anterior-lateral side, near the stem tip.

Loading consisted of three cycles of loading-unloading between 10 N and 3000 N of axial compression force, parallel to the long axis of the tibia at a rate of 300 N/s. We did not observe any significant changes in the force-displacement profile or surface strains among the three cycles. The forces were applied at 70% of the AP implant dimension and were offset 3.3 mm medially to the center of the medial condyle (Fig. 4). This loading condition is based on Rawlinson et al. [17] and represents a critical scenario of four times body weight acting on a single condyle with an additional 10 Nm varus moment, which is the largest acceptable moment for a constrained condylar knee, based on the strength limit of the polyethylene post. The first two cycles of loading for each specimen were considered preliminary to settle the specimen in the test frame; data from the third cycle were used to validate the finite-element model.

Fig. 4.

Location of load application for the validation.

Finite-Element Modeling

The 3D geometries of the nine tibiae were reconstructed from the preimplantation CT scans. The reconstructed bones included the resection, the created defect, and, for the stem groups, the reamed canal geometries. We considered 220 mm of the tibiae distal to the resection for the models. The geometries of the tibial implant components were reconstructed from 3D laser scans with an accuracy of 40 µm (Range 7 3D Scanner, Konica Minolta Inc) of pristine components using reverse-engineering software (Design X, 3D Systems Inc). To determine the implant position within the bone, we performed a best-fit alignment between the distal end of the tibiae reconstructed from the preimplantation and postimplantation CT scans. The distal end of the tibiae had no metal artifact, which allowed for accurate reconstruction and subsequent best fit alignment of the bony geometry. This technique has been used previously to determine component alignment [16, 22]. For the validation of the finite-element models, we also included the distal potting and the load application fixture. We chose not to virtually generate all three implantations for all specimens, but to only model the experimental implantation for each specimen to minimize the assumptions for component position and cementing technique when creating the models.

The models were meshed with linear tetrahedral elements in Abaqus 2017 (Dassault Systèmes). The element size increased from 1 mm at the bone-implant interface edges to 1.5 mm at 40 mm distal to the resection to 2.5 mm at 80 mm distal to the resection and to 3.5 mm at 200 mm distal to the resection. This resulted in between 1.1 and 1.6 million elements for each model.

All materials were modeled as linear and isotropic. The implant components were modeled as homogeneous Ti6Al4V alloy with an elastic modulus (E) of 114 GPa and Poisson ratio (v) of 0.33. The bone was modeled as nonhomogeneous with material properties derived from the preimplantation CT scans using empirical relationships between the apparent density of bone and E. To this end, we interpolated the Hounsfield units (HUs) to the finite-element mesh. Then, we converted the HUs for each element to bone mineral density using the calcium hydroxyapatite bone density reference phantoms included in the CT scans. The bone mineral density was converted to ash density and subsequently to apparent density according to the relationships proposed by Schileo et al. [19]. Finally, we obtained the elastic modulus of bone from its apparent density with the relationships proposed by Morgan et al. [12] for bone densities below 0.41 g/cm³, Snyder and Schneider [21] for bone densities above 1.75 g/cm³, and with linear interpolation for densities between 0.41 g/cm³ and 1.75 g/cm³. This combination of density-modulus relationships led to physiological strains for the tibia during daily activities [25], with good correlation with experimental data [24]. To determine the extent of the cement mantle, we used the information from the CT scans of the specimens with the created defect to identify the elements that were not bone or marrow. These were assumed to be cement and were modeled as homogeneous polymethylmethacrylate (PMMA) with E = 2.2 GPa and v = 0.3. When included, the base potting of the distal tibiae was modeled as homogeneous PMMA (E = 2.2 GPa, v = 0.3), and the load application fixture was modeled as homogeneous aluminum alloy (E = 70 GPa, v = 0.3).

We modeled the cement-bone interface as tied (that is, no relative displacement allowed across the interface), corresponding to perfect cement-bone interdigitation. We modeled the interaction between the implant components and the cement through zero-thickness cohesive elements with a tensile stiffness of 57.3 MPa/mm and a shear stiffness of 151.4 MPa/mm [6]. We modeled the contact between the implant surfaces and the bone as a surface-to-surface frictional contact with a friction coefficient of 0.6 for the external surface of the cone and 0.2 for the surface of the long, diaphyseal-engaging stems.

Finite-Element Model Validation

A first set of simulations was intended to validate the finite-element models against the experimental results and included the load application fixture, the implanted tibia, and the distal potting to replicate the experimental setup (Fig. 3B). In this set of simulations, we constrained all degrees of freedom of the potting fixture and applied 3000 N to the loading fixture to simulate the loading experiment (Fig. 3B). The exact position of the loading fixture was obtained by the set of retroreflective markers attached to the implanted bone and the loading fixture (Fig. 3A). The main outcome of this model was the strains at the locations of the strain gauges, which we compared against the experimental data from the strain gauges to validate that the finite-element model was simulating the experiment. We performed a regression analysis to quantify the correlation between experimental and computational data. We quantified the overall agreement between the computational models and the experimental results using the root mean square error (RMSE), computed as shown in Equation 1:

$$RMSE=\sqrt{\frac{{{\displaystyle \sum }}^{\text{}}{\left({\epsilon }_{exp}-{\epsilon }_{comp}\right)}^2}{N}}$$ (1)

where Σ is the sum operator, N is the number of points where the strain was measured across all specimens, and ε_exp and ε_comp are the experimental and computational strains, respectively.

Simulation of Stair Ascent

A second set of simulations was intended to evaluate the biomechanical advantage of cones and stems during the demanding daily activity of stair ascent. In this set of simulations, we did not include the potting or the load fixture. We discretized the knee forces and moments measured experimentally during stair ascent [4] into 26 loading steps. These joint loads were scaled by the body weights of the donors from whom the specimens were obtained (Table 1). The loads were directly applied to the proximal surface of the baseplate, and all degrees of freedom of the distal end of the bones were fully constrained.

We evaluated the risk of debonding during stair ascent by using the interfacial normal and shear stresses to compute the modified multiaxial Hoffman failure criterion proposed by Zelle et al. [30] for the strength of the implant-cement interface under mixed-mode tensile and shear loading conditions. This criterion was experimentally derived from in vitro tests [30] and has been used in computational models to evaluate the risk of debonding between implant and cement in TKA [6, 26, 31].

$$FI=\frac{{\sigma }_s}{S_s}+\frac{{\sigma }_n}{S_t}if {\sigma }_n\ge 0$$ (2)

$$FI=\frac{{\sigma }_n^2}{S_t{S}_c}+\left(\frac{1}{S_t}-\frac{1}{S_c}\right){\sigma }_n+\frac{{\sigma }_s^2}{S_s^2}if {\sigma }_n<0$$ (3)

Equations 2 and 3 are used according to the value of the normal (σ_n) and shear (σ_s) interfacial stresses and the interfacial strengths in shear (S_s = 3.89 MPa), tensile (S_t = 2.09 MPa), and compressive (S_c = 70 MPa) directions [27, 30] to compute the failure index (FI) for tensile and compressive interfacial stresses, respectively. Values of the failure index ≥ 1 represent interface failure.

We also evaluated the relative micromotion at the interface between the bone and the external face of the cone during stair ascent. This cone surface is intended for ingrowth, so we compared the differences in displacement between each pair of closest bone-implant nodes to the most restrictive threshold for bone ingrowth of 20 µm or less of micromotion across the interface [8].

For each outcome variable, micromotion and failure index, we calculated the composite results of the entire cycle as the maximum value of the variable at each node or element of the model throughout the entire cycle [23]. This represents the burden placed on the model during each cycle of the simulated activity.

Ethical Approval

Ethical approval for this study was waived by the institutional review board at the Hospital for Special Surgery (protocol # 2016-0071-CR4).

Results

Experimental Testing and Validation of the Finite-Element Models

In two specimens, the strain gauges on the anterior side of the bone were inadequately glued to the bone surfaces leading to erroneous measurements of strain, which was only noticed during data processing. For all other strain measures, the tensile and compressive principal strains obtained computationally on the anterior-lateral and posterior-medial surfaces of the tibia, respectively, matched the experimentally measured strain values at the same locations (R² = 0.84) (Fig. 5). However, one specimen with a metaphyseal cone filling the defect and a long, diaphyseal-engaging stem had particularly poor agreement between computational and experimental results for the posterior strain gauge, with a difference of 1025 µε. This was double the second largest difference in strain, which was 552 µε. Despite the relatively large error for this specimen, the root mean square error in strain was 398 µε.

Fig. 5.

Results from the validation in which the absolute value of the experimental principal strain measured by the strain gauges on the surface of a particular tibia are plotted against the absolute value of the principal strains calculated at the same location on the surface using the finite-element model of the same bone. The dotted line corresponds to the case where the computational results are equal to the experimental data, representing the ideal agreement. Diamond markers correspond to the tensile principal strain from the gauges located at the anterior-lateral surface of the bone, at the stem tip; circular markers correspond to the compressive principal strain from the gauges located at the posterior-medial surface of the bone, in line with the loading application fixture; FE = finite element. A color image accompanies the online version of this article.

Risk of Cement-implant Debonding During Stair Ascent Predicted by the Finite-Element Models

All configurations had similar risk of interfacial failure, with no portion of the cement-implant interface at risk of debonding during stair ascent, as demonstrated by composite interfacial failure indexes below the failure limit of 1 (Fig. 6). Considering the entire interface, the maximum composite values of failure index during stair ascent (Fig. 6) for the group with no cone and a short, cemented stem (0.74, 0.59, and 0.35) were similar to the group with a cone and a short, cemented stem (0.68, 0.57, and 0.37). The failure index values for the group with a cone and a long, diaphyseal-engaging stem (0.22, 0.3, and 0.18) were smaller than when the cone was combined with a short, cemented stem (Fig. 6). However, when considering the portion of cement-implant interface common to all three configurations (such as the most proximal 35 mm of the interface), the failure index values for the no cone with a short, cemented stem group (0.33, 0.25, and 0.19) were larger than for the cone with a short, cemented stem group (0.22, 0.17, and 0.17), which were like those for the cone with a long, diaphyseal-engaging stem group (Fig. 6).

Fig. 6.

A-B Peak failure index at the cement-implant interface for (A) the most proximal 35 mm and (B) around the short, cemented stem.

For all specimens and configurations, most of the cement-implant interface had minimal risk of debonding, and the differences across configurations were small. In this way, the cement-implant interface area with failure index less than 0.1 was 77%, 79%, and 86% for the specimens implanted with no cone and a short, cemented stem; 89%, 87%, and 93% for the specimens implanted with a cone and a short, cemented stem; and 92%, 92%, and 94% for the specimens with a long, diaphyseal-engaging stem. In all instances, the largest failure index values were concentrated in the distal and posterior zones of the cement-implant interface (Fig. 7). For the groups with short, cemented stems, these zones with the greatest failure index were around the distal areas of the stem, where no cement was present for the long, diaphyseal-engaging stem group.

Fig. 7.

Distribution of failure index at the interface between the implant and the cement.

Cone-bone Micromotion During Stair Ascent Predicted by the Finite-Element Models

The differences in cone-bone micromotion between the specimens with short, cemented stem and those with long, diaphyseal-engaging stem were small. The composite peak micromotions between the cone and the bone were 28 µm, 31 µm, and 11 µm for the specimens implanted with a short, cemented stem, and 23 µm, 15 µm, and 13 µm for the specimens that received a long, diaphyseal-engaging stem (Fig. 8). For all specimens and configurations, the largest micromotion was concentrated at the posterior aspect of the cone and most, if not all, of the interface area between the cone and the bone had micromotions below the most restrictive threshold for bone ingrowth of 20 µm (Fig. 9).

Fig. 8.

Peak micromotion at the interface between the cone and the bone intended for ingrowth.

Fig. 9.

Distribution of micromotion at the interface between the cone and the bone intended for ingrowth.

Discussion

Cones and stems are commonly used in revision TKA to address moderate metaphyseal defects and achieve long-term implant fixation. Despite their excellent clinical outcomes, some studies report no difference in outcomes for cones over cement [5] and for long over short stems when using cones [7]. In this way, the biomechanical evidence is limited, and prior biomechanical studies have reported conflicting results about the efficacy of long stems [2, 13, 28]. Therefore, it is still unclear whether cones provide any biomechanical advantage over cementing alone and what is the biomechanical efficacy of long stems when combined with cones to address metaphyseal defects in revision TKA. We combined experimental cadaveric testing with finite-element modeling to evaluate the biomechanics of cones and stems for addressing moderate tibial defects. We found that although metaphyseal cones helped reduce the risk of debonding at the cement-implant interface during the demanding activity of stair ascent, the differences were small, and cement alone can provide comparable results to cones in contained metaphyseal defects. However, when using a cone, we did not find a biomechanical advantage of long, diaphyseal-engaging stems over short, cemented stems.

Limitations

Our study has limitations. First, this was a biomechanical study with well-controlled implantation and loading. In this way, uncertainty exists in how the observed differences in the outcome variables will translate to clinical practice; however, the study design allowed us to understand the basic biomechanical principles behind the use of cones and stems. Moreover, our conclusions agree with clinical studies that found no superiority of cones [5] and that short stems are an effective adjunct when used with cones [7]. Second, we approximated the metaphyseal defect by a conical cavity that was reamed in the bone, and we assumed perfect line-to-line fit between the metaphyseal cone, when used, and the bone. This situation represents a best-case scenario for the groups including a cone due to the integrity of the defect walls and the relatively large defect size to be filled with cement alone. However, we could use this approach to control the type and size of defects across the different specimens of each set, allowing us to compare the biomechanical behavior of the different implantation approaches independently of the created defect. Third, to simulate stair ascent, we considered the joint loads and moments measured by Bergmann et al. [4] using in vivo instrumented primary TKA components. Although stair ascent is a demanding daily activity, primary TKA components are often less constrained than revision TKA components, especially in varus-valgus. Therefore, the applied loads likely underrepresent the joint moments, particularly the varus-valgus moment. This may have contributed to the small differences among groups. However, to our knowledge, no equivalent loading data exist for more constrained implants. Future studies could evaluate the sensitivity of the different reconstruction techniques to other demanding loads from daily activities and to increased varus-valgus loads. Fourth, although we considered nine specimens, this number was insufficient to capture sufficient interspecimen variability. In this way, in addition to the specimen variability, we observed variability in implantation of the cone, stem, and baseplate, which may have confounded some of our results. We felt it was important to use a combination of finite-element modeling and experimental validation rather than conduct experimental tests of a larger number of cadaveric specimens. Moreover, although we could have virtually re-created all implantations in all specimens to increase the sample size, the number of specimens would still be small, and we believed it was more important to reduce the modeling assumptions in terms of component position and cementing technique. Nonetheless, follow-up studies with more specimens are needed to better capture interspecimen and implantation variability. Fifth, we only simulated the immediate postoperative scenario and did not account for progressive changes due to repetitive loads (such as bone collapse or debonding progression). Future studies should examine these cyclic loading conditions. Sixth, although the outcomes of interest of the study were the micromotion between the cone and the bone and the cement-implant debonding, we performed the validation based on the surface strains. Unfortunately, direct validation of the outcomes of interest was not possible because the involved surfaces were not directly visible or accessible. Nonetheless, bone surface strains have previously been used to validate finite-element models to study bone-implant interaction [24]. Seventh, we did not account for the bone anisotropy in the finite-element models. Although the bone is indeed anisotropic, considering the bone as linear isotropic is a common modeling practice in finite-element models [2, 13, 15-17, 23-25] and has been shown to yield similar results to anisotropic bone behavior [14].

Risk of Cement-implant Debonding During Stair Ascent Predicted by the Finite-Element Models

We found that although no point along the implant-cement interface was at risk of debonding for any configuration during stair ascent (that is, failure index < 1) and the differences among configurations were small, adding a cone to a short, cemented stem improved the distribution of failure index in all cases, resulting in a greater proportion of the cement-implant interfacial area with less risk of debonding, particularly in the most proximal 35 mm of the cement mantle. In this way, using a cone reduced the peak failure index value in all instances except around the short, cemented stem of one specimen set. However, even for this specimen set, a cone still improved the overall distribution of failure index around the stem. To our knowledge, no prior biomechanical study has compared the effectiveness of metaphyseal cones against cement alone for addressing moderate metaphyseal defects. Nonetheless, our results agree with clinical reports of excellent survivorship of metaphyseal cones. Moreover, the small differences found between using a cone or cement alone are in line with one study that reported no superior clinical outcomes for metaphyseal cones over cementing alone [5]. In this way, although our results support using cones for addressing moderate metaphyseal defects, the small differences found suggest that using cement alone to fill the defect may be sufficient for some situations. However, it is unclear whether cementing alone would still perform comparably to using a cone in the long term under repetitive loading.

When using a cone, a long, diaphyseal-engaging stem did not improve the overall distribution of failure index in the most proximal 35 mm of the cement mantle compared with the distribution for a short stem. In this way, a long, diaphyseal-engaging stem reduced the peak failure index in the most proximal 35 mm of the cement mantle compared with a short stem only in one specimen set. Moreover, for one specimen set, a long, diaphyseal-engaging stem had the largest peak failure index of all three configurations in the most proximal 35 mm of the mantle. For this set, the peak failure index values with a long, diaphyseal-engaging stem were concentrated at the distal anterior edge of the cement mantle, where the mantle was thin due to the small space between the implant and the cone. Long stems have been shown to be effective in reducing the load at the metaphyseal bone [17]. However, their usefulness when using metaphyseal cones remained unclear. Our results suggest that when a metaphyseal cone is used, long stems do not provide a biomechanical advantage over short stems. Similarly, a prior biomechanical study found no superiority of long stems over short stems when using cones for addressing metaphyseal defects, even if the defect is uncontained [28]. Moreover, our results coincide with Jacquet et al. [7], who performed a retrospective review of revision TKA with a rotating hinge implant at an average 9.3-year follow-up and reported that, when using trabecular metal cones to address metaphyseal defects, short stems had no difference in survivorship and better patient-reported outcomes than long stems.

Cone-bone Micromotion During Stair Ascent Predicted by the Finite-Element Models

We also found that when using metaphyseal cones, both short and long stems resulted in a stable cone-bone interface compatible with bone ingrowth. In this way, we found that the peak micromotions between the cone and the bone were greater when using a short, cemented stem than when using a long, diaphyseal-engaging stem for all but one of the specimen sets. However, in all specimen sets, most of the cone-bone surface had micromotion values less than 20 µm, the most conservative threshold for ingrowth [8]. Thus, short and long stems appear to provide sufficient stability at the cone-bone surface to allow bone ingrowth. Similarly, a clinical study reported no tibial aseptic loosening when metaphyseal cones were combined with short or long stems, evidencing their ability to achieve stable fixation by bone ingrowth [7]. Moreover, our results with cadaveric-based models confirm the results of prior biomechanical studies that found, using models based on synthetic bone, small decreases in sleeve-bone and cone-bone interfaces when using long stems [13, 28]. Furthermore, our values of peak micromotion (11 µm to 31 µm) are similar to the values reported by Xie et al. [28] and Awadalla et al. [2]. These studies show that, despite their design differences, cones and sleeves have similar biomechanical behavior.

Conclusion

Although metaphyseal cones help reduce the risk of implant-cement debonding, no point in the cement-implant interface was at risk of debonding for any configuration. Moreover, the differences with filling metaphyseal defects with cement alone were small, suggesting that although cones reduce the biomechanical burden at the cement-implant interface, cement alone may be sufficient for moderate, contained metaphyseal defects, even though studies that consider the long-term behavior are needed. Further, when using cones to address moderate tibial metaphyseal defects in revision TKA, we could not show a clear biomechanical advantage of long, diaphyseal-engaging stems over short, fully cemented stems. This is the first study to examine the biomechanics of cones with computational models constructed from multiple experimentally implanted cadaveric specimens. Future studies should explore additional loading conditions, quantify the interspecimen variability, and evaluate the behavior of the reconstructive techniques under repetitive loads.