Detection of simulated complications in all-polymer PEEK knee prostheses: a cadaveric imaging study using Radiography, CT, and MRI

Abstract

Background

All-polymer polyetheretherketone (PEEK) knee prostheses are being developed as an alternative to metal implants, offering benefits like bone-like elasticity, wear resistance, and the absence of clinically relevant metal artefacts on imaging. Their radiological characteristics and complications have not been systematically studied. This study describes the imaging features of a new all-polymer PEEK knee prosthesis and assesses the detection of simulated complications with radiography, computed tomography (CT), and magnetic resonance imaging (MRI) in cadaver models.

Methods

The study followed Article 68, paragraph 1 of the Dutch Civil Code for the lawful use of human cadaver material, and all procedures met institutional and national ethical guidelines. Four cadaveric lower limbs were implanted with cemented PEEK knee prostheses (PEEK femoral and polyethylene tibial components). Radiography, CT, and MRI were used to evaluate both typical postoperative findings and four simulated complications: prosthesis loosening, cement fracture, implant migration, and prosthesis fracture. An experienced fellowship-trained musculoskeletal radiologist assessed the visibility and diagnostic characteristics of each complication.

Results

The prostheses were clearly visible across all imaging modalities, allowing reliable assessment of the implant–cement–bone interface without relevant metal artefacts. Cement fractures were detected on all radiographs, though visibility varied by projection. CT distinguished PEEK and polyethylene based on radiodensity and provided the most complete structural overview of all simulated complications. T2-weighted MRI was the most sensitive modality for detecting fluid at the implant–cement interface and identifying interface separation.

Conclusions

All-polymer PEEK knee prostheses demonstrate favourable imaging characteristics in a cadaveric setting, without any clinically relevant metal-related artefacts and with good visibility of simulated implant complications across radiography, CT, and MRI. These findings suggest potential imaging advantages for future clinical follow-up of polymer-based implant systems, although direct comparative clinical studies are required for confirmation.

Background

Polyetheretherketone (PEEK) is a promising alternative material for total knee arthroplasty (TKA) implants due to its high wear resistance, bone-like elasticity (3.6–4.0 gigapascals), and radiolucency (PEEK: 188 HU) [1]. Due to its unique properties, PEEK implants appear differently from standard TKA in all types of standard medical imaging modalities. This requires training to familiarize radiologists with the different appearances of PEEK TKA components in medical images. However, a metal-free TKA may enable improved post-operative monitoring due to the absence of metal artefacts. This means events or complications that are difficult to detect with metal TKA implants may be visualized in PEEK TKA. However, knowledge of the radiological performance or appearance of such complications is limited. This study, therefore, aimed to investigate the radiologic appearance of potential complications in metal-free TKA reconstructions, which were simulated in cadaveric knees using a cemented PEEK femoral TKA component and polyethylene (PE) tibial component. For this purpose, complications with increasing severity were simulated, ranging from implant loosening [2], cement fracture [3], implant migration [4] to implant fracture [5]. These failure modes were subsequently imaged using radiography (XR), computed tomography (CT), and magnetic resonance imaging (MRI). An experienced musculoskeletal fellowship-trained radiology consultant evaluated the images to provide a preliminary framework for future clinical assessment.

Materials and methods

This was an experimental imaging study (Fig. 1) using human cadaveric lower limbs. The study was conducted in accordance with Article 68, Paragraph 1 of the Dutch Civil Code concerning the lawful delivery and use of human cadaveric material. All procedures complied with both institutional and national ethical standards governing cadaveric research. No institutional review board (IRB) approval was required, as no living participants were involved.

Fig. 1.

Flowchart illustrating the experimental study design and imaging workflow

Four cadaveric lower-limb specimens were implanted with cemented all-polymer knee prostheses (PEEK femoral component and a PE tibial component) according to the surgical instructions by a staff orthopedic surgeon with more than 10 years of experience in knee arthroplasty at an academic hospital. The femoral and tibial components were fixated using Palacos R bone cement mixed under a vacuum using the Palamix system (Heraeus Medical GmbH, Wehrheim, Germany). Medical imaging was performed under five sequential conditions of the femoral component: the intact condition (directly after placement of the implants), followed by four adverse events, including prosthesis loosening, cement fracture, migration, and prosthesis fracture. After each condition, the joint cavity was sutured and filled with a controlled volume of saline to approximate physiologic intra-articular fluid levels and reduce intra-articular air, after which the imaging protocol was implemented.

Implant loosening was simulated by gently loosening the femoral implant from the cement mantle using the impactor tool and a chisel. The impactor was used as a leverage tool to flex the femoral component, while the chisel was used to lift the anterior flange. Subsequently, the posterior condyles were loosened from the cement, also using the chisel. This model reflects late-stage mechanical loosening with clear interface separation. It does not replicate the initial biologically driven phases of loosening seen in vivo, such as wear-particle–induced inflammation, osteolysis, marrow edema, or subtle interface debonding, which are challenging to reproduce experimentally in cadaver studies.

Cement mantle fractures were created by completely removing the femoral implants. Upon implant removal, cement would typically stick to the femoral component, thereby forming cement cracks, particularly in the medial and lateral distal condyles.

Implant migration was simulated by creating a small gap between the implant and the bone. To ensure implant migration was maintained after joint reduction, small pieces of a wooden spatula (thickness ~ 1 mm) were positioned between the implant and the cement at the distal condyles.

Finally, an implant fracture was simulated by manually fracturing the femoral component after removal. For this purpose, the anterior flange was fixated in a vice, after which the posterior condyles were moved posteriorly with a clamp until fracture occurred. This generally resulted in two fractures per condyle, with the fractured component consisting of the anterior flange, two distal condyle fragments, and two posterior condyle fragments. The fragments were rejoined using tape and placed back into the joint cavity. Implant fractures very rarely occur in TKA, and preclinical testing of the current implant system indicated that fractures are highly unlikely. Nonetheless, considering the novelty of the implant material in TKA, the aim was to investigate the radiological appearance of even such an unlikely event.

Imaging was conducted across three modalities: radiography (XR), computed tomography (CT), and magnetic resonance imaging (MRI). Standardized imaging protocols were applied to capture axial, sagittal, and coronal views in each modality. These baseline images served as controls for later comparisons. At each complication stage, the specimens were re-imaged using the same protocols to maintain consistency:

Radiography (Siemens Ysio)

AP view: 60 kV, 8 mAs.

Lateral view: 60 kV, 8 mAs.

Skyline View (merchant view): 70 kV, 8 mAs.

CT scan (Siemens Sensation)

Scan Parameters: 1 mm collimation, 0.75 mm slice thickness, 120 kV, 100 mAs.

Non-contrast.

Reconstructions: Axial, sagittal, and coronal views using a bone kernel.

MRI (Siemens Avanto 1.5T)

T1 nFS TSE nFS Axial + Coronal (Slice Thickness / Gaps: 3 mm / 0.3 mm).

PDFS TSE Axial + Coronal (Slice Thickness / Gaps: 3 mm / 0.3 mm).

PD nFS TSE Sagittal (Slice Thickness / Gaps: 3 mm / 0.3 mm).

T2FS TSE Sagittal (Slice Thickness / Gaps: 3 mm / 0.3 mm).

(TSE = turbo spin-echo; PD = proton density; FS = fat-saturated; nFS = non–fat-saturated)

Short Tau Inversion Recovery (STIR) fat saturation sequences, commonly utilised due to their reduced sensitivity to metal artefacts in post-operative imaging, were not employed in this instance. Instead, frequency-selective fat-saturated PD sequences were chosen for their higher spatial resolution, and because no metal was present in the PEEK prosthesis.

All imaging series were reviewed by a board-certified radiologist who completed a dedicated musculoskeletal radiology fellowship and has 6 years of experience in musculoskeletal imaging to evaluate the appearance of each complication type on radiography, CT, and MRI. Image interpretation was performed by a single reader. No formal interobserver analysis, repeat reading, or internal consensus reading was conducted. This comparative assessment aimed to validate findings across modalities and identify radiological features characteristic of each simulated failure mode.

Descriptive analysis was used; no formal statistical testing or power calculation was performed, given the experimental nature and limited number of specimens.

Results

Intact implants

Across all four specimens, radiography, CT, and MRI enabled clear visualization of the PEEK knee prosthesis positioning and bone-cement interface (Fig. 2). Common artefacts, such as gas in soft tissues and joint spaces, as well as diffuse soft tissue edema, were noted in all modalities and are consistent with cadaveric specimen conditions. All imaging modalities confirmed correct alignment, with no evidence of implant fracture or migration.

Fig. 2.

An intact PEEK implant is shown using radiography (A), CT (B), and MRI (C) lateral views. The femoral PEEK (white arrow) and tibial PE (brace) components, as well as the cement layer (long thin arrow), are identified. Additionally, both intra- and peri-articular gas presence is noted (star)

1. Radiography

   On the radiography, the femoral PEEK component is completely radiolucent with a radiodensity equal to the surrounding soft tissues (Fig. 2). The contour of the femoral component is indirectly apparent because of the surrounding tissues or gas providing contrast (such as bone, fluid, or cement). The cement layer is clearly depicted, as well as the tibial radiolucent (PE-material) insert.

2. Computed Tomography (CT):

   CT consistently provides a detailed view of the bone-cement interface, which is often only partially visible in metallic knee replacements due to metal beam hardening artefacts.

   CT imaging enables clear differentiation between polyethylene (PE) and polyether ether ketone (PEEK) components due to their distinct radiodensity differences (Table 1) (Fig. 2). These values reflect the denser molecular structure of PEEK compared to PE.

3. MRI.

   Small amounts of fluid were observed in the cement-bone interface in either or both the femoral and tibial components on MRI.

   On MRI, the cement layer appears completely hypointense due to its very short T2 relaxation time, which prevents signal generation. This hypointensity is similar to that observed in the femoral PEEK and tibial PE components (Fig. 2). Consequently, the cement layer could only be distinguished from the PEEK and PE components in areas where fluid was present at the interface. Therefore, MRI is limited in detecting cement issues such as spillage or fractures.

   Scattered micrometallic magnetic susceptibility artefacts were observed, likely resulting from residual metallic particles left behind by surgical instruments. These artefacts did not have a significant effect on the evaluation of any structures, as they did not cause any relevant obscuration of anatomical details.

Table 1.

Hounsfield units (HU) values of the implant components

Component	Mean HU	Median HU	Max HU	Min HU
Femoral PEEK Component	187.6	187	223	142
Tibial PE Component	−79.2	−79	−47	−109
Cortical Bone	–	–	> 1000	800
Cancellous Bone	–	–	800	300
Palacos Cement	1328	1328	1460	1195

Table legend: Reference Hounsfield unit ranges for cortical and cancellous bone were derived from Radiopaedia [21]

Reference Hounsfield unit values for PMMA bone cement (e.g., Palacos) were derived from published CT imaging literature [22]

Cement fractures

1. Radiography.

   The cement fractures were detected on all radiographs and CT scans.

   They were identified as either a focal indentation (1 out of 4) or fine lucent interruptions and fracture lines in the cement layer (3 out of 4) (Fig. 3). The full extent of the fracture lines could not be observed entirely on a radiograph, unlike CT, which did reveal the complete extent.

   The cement fractures were detectable in 2 out of 4 cases on lateral views, and in both lateral and axial views (Fig. 3) in the other 2 cases. No cement fractures were identified on coronal views.

2. Computed Tomography (CT).

   CT imaging enabled the clear identification of all fracture lines and their complete extent (Fig. 4). In each case, gas and/or joint fluid present in the fracture lines facilitated their detection.

   In one case with multiple cement fracture lines, one of the fracture lines pre-existed in the initial intact study, likely due to the preparation technique.

3. MRI.

   MRI visualized the cement fractures in three out of four cases, with the fourth showing subchondral gas indicative of a fracture site.

   MRI was sometimes limited in visualizing the full extent of cement fractures if parts of the undisplaced narrow fracture sites were not filled with joint fluid or gas to provide sufficient contrast in relation to the hypointense properties of cement.

   In one instance, the fracture was not directly visible, but subchondral gas at the anterior aspect of the medial femoral condyle indicated its location, consistent with corresponding CT findings. In the other three instances, the cement fractures were apparent on all sequences, but the PD nFS sequence offered the best delineation.

Fig. 3.

Cement fractures. The disruption of the cement layer at the medial femoral trochlea (circle) is illustrated in the radiography (A) and CT (B) axial view. Additionally, another cement fracture is noted at the lateral trochlea, showing significant loosening of the cement layer (arrows), accompanied by interspersed gas (CT) and fluid (MRI) at the interface

Fig. 4.

Horizontal cement fracture (arrows) along the femoral trochlea visible as a thin disruption on the sagittal CT image (A). The complete extent of the cement fracture is most clearly shown on the coronal CT image

On MRI, the cement fractures were visible as either small disruptions in the cement layer (Fig. 3) or larger fluid-filled gaps. One case showed a fracture as a cement layer indentation at the medial femoral trochlea. In another case, subchondral bone marrow edema was apparent at one of its cement fracture sites at the lateral femoral trochlea.

In each of the four instances, a minor accumulation of fluid was observed at the prosthesis-cement interface. This occurred due to the temporary removal of the prosthesis, which was part of the preparation technique used to simulate cement fractures.

Implant loosening and migration

In general, observations in imaging of implant loosening and implant migration cases were quite similar, with implant migration further enlarging the observed effects. As such, the observations of these two simulated adverse events are discussed here simultaneously.

1. Radiography.

   All four radiographic studies revealed implant loosening as a subtly increased prosthesis-cement distance, indirectly visible as a fine line of interspersed gas in the interface (Fig. 5). These findings were apparent on all lateral views, but only in 3 out of 4 axial patellar views, and only one AP-view.

2. Computed Tomography (CT).

   Implant loosening and migration was evident on all CT scans as (mild) widening of the prosthesis-cement interface, aided by easy-to-detect interspersed air and, to a lesser extent, joint fluid (Fig. 5). Although true migration requires assessment over multiple time points, in our cadaveric model this was simulated by creating a small gap at the interface, resulting in a more pronounced separation compared to simple loosening.

   On CT, in one instance, the increased amount of joint fluid in the widened interface (instead of gas) resulted in reduced image contrast. Consequently, in this case, adjusting the image windowing (also known as histogram modification) towards a soft tissue window improved visualization.

3. MRI.

   Signs of loosening and migration were most readily detected on MRI, where fluid accumulation in the widened implant-cement interface created a high signal contrast (Fig. 5).

   Implant migration was evident in all cases as it involved a more noticeable widening of the prosthesis-cement interface, permitting a greater inflow of air and joint fluid.

Fig. 5.

Implant loosening can be identified by the widening of the implant-cement interface (arrows). In radiography (A), this appears as a gas-filled gap at the widened interface, comparable to the CT image (B). The fluid present in the widened interface is most prominently visible on MRI (C)

Implant fractures

1. Radiography.

   Radiographically, implant fractures were detected in three out of four cases. These were visible solely on lateral projections as a disruption in the femoral implant component, indirectly noticeable due to gas, and to a lesser extent, fluid filling the disruption (Fig. 6). In one case, the implant fracture at the trochlear aspect was radiographically occult.

2. Computed Tomography (CT).

   CT scans did provide clear identification of all implant fractures, all located at the trochlear aspect of the femoral implant. Three cases exhibited fine disruptions filled mainly with gas, while one case displayed a wider disruption (up to 3 mm diastasis) containing both fluid and gas. Another case demonstrated multiple crisscrossed gas-filled fracture lines located at the trochlear aspect of the implant.

   Enhancing gray-level windowing for soft tissues facilitated the identification of fluid-filled implant fracture lines due to improved contrast differentiation between fluid (0–30 HU) and the implant PEEK (average: 188 HU) polymer (Fig. 6).

3. MRI.

   MRI enabled the visualization of all implant fracture lines, as these lines were made visible due to interspersed fluid, providing high T2 hyperintense signal contrast, especially on PDFS and T2FS sequences (Fig. 6). Fracture lines were detectable across all sequences, including non-fatsat T1 weighted sequence, where the disruption in the hypointense signal of the implant material indicated the presence of fractures.

   Certain segments of the fractures are partially obscured due to gas-filled areas within the fracture lines, resulting in dark signal voids that artificially fill the fracture line.

Fig. 6.

Implant fracture at the femoral trochlea (highlighted) appears as a faint gas-filled gap on radiographic imaging (A). Both MRI (PD nFS) (B) and CT (C) scans show that the fracture is also filled with fluid. Modifying the window level to soft tissue settings on CT enhances the visibility of the fracture lines due to improved contrast (D)

Detection rates for each simulated complication across imaging modalities are summarized in Table 2.

Table 2.

Detection rates of simulated implant complications across imaging modalities

Simulated Complication	Radiography	CT	MRI
Cement fracture	4/4	4/4	3/4
Implant loosening / Migration	4/4	4/4	4/4
Implant fracture	3/4	4/4	4/4

Values represent the number of specimens in which the complication was detected (n = 4). All imaging studies were interpreted by a single fellowship-trained musculoskeletal radiologist

Discussion

This study was conducted to evaluate the effectiveness of XR, CT, and MRI in visualizing PEEK knee prosthesis positioning and identifying simulated implant complications in a controlled cadaveric model. PEEK implants are being considered for knee replacements due to their great wear resistance (PEEK OPTIMA wear factor: 0.0384 × 10⁻⁶ mm³/N·m), elastic modulus closer to cortical bone (–4 GPa), reflecting a relatively low flexural modulus, and low radiopacity, which is beneficial for post-surgery radiologic follow-up [6, 7]. However, because PEEK is a relatively new implant material, there is limited evidence about how well it performs in radiological evaluations.

Our study findings demonstrated that CT provided the most complete structural assessment of the prosthesis–cement–bone interface and the full extent of all simulated complications (cement/implant fractures and implant loosening/migration).

The loosening and migration models in this study reflect advanced mechanical failure, not early implant loosening seen clinically. In vivo, loosening usually begins with biological processes such as wear-particle inflammation, bone marrow changes, and progressive osteolysis, which initially present as subtle imaging abnormalities rather than clear interface gaps. Our cadaveric model used mechanical disruption, resulting in more obvious separation, and therefore represents later-stage failure. Thus, our findings on detectability may not apply to early biologically driven loosening.

Although MRI was less effective in evaluating cement conditions, it successfully identified all instances of implant loosening, simulated migration, and implant fractures due to its high sensitivity to fluid-based changes and detection of even minimal amounts of interspersed joint fluid.

In the context of a metal-free PEEK TKA, MRI has strong potential to emerge as the most sensitive modality for detecting early loosening-related changes, which are more likely to manifest as bone marrow edema, small fluid collections, and early osteolytic defects rather than obvious structural discontinuities.

CT remains the most complete modality for defining established structural damage and for preoperative planning, but it may be less sensitive than MRI for detecting the earliest fluid-rich and marrow-based changes. These complementary roles of CT and MRI are likely to be relevant in the follow-up of polymer-based implant systems.

Lyu et al.‘s recent clinical study demonstrates that MRI is useful in assessing PEEK-based total knee arthroplasty postoperatively. Their 3D MRI findings correlate knee effusion with inflammatory markers, functional outcomes, and range of motion, supporting MRI’s role in detecting abnormal inflammation, persistent effusion, and early loosening in metal-free implants [8].

In contrast, the detection of loosening in routine metallic knee implants can be challenging on both MRI and CT due to metal artefacts, particularly from cobalt-chrome components, which generate pronounced susceptibility artefacts that obscure the implant-cement interface [9]. Nevertheless, studies such as Endo et al. have shown that MRI can achieve high sensitivity and specificity for certain components (e.g., oxidized zirconium), while its performance is reduced for cobalt-chrome components [10].

In our study, subchondral bone edema near the cement fracture site was observed on MRI as an indirect indicator in one case, potentially due to mechanical impact during preparation. The presence of reactive bone marrow edema in clinical scenarios remains uncertain.

While radiography is overall useful, it often struggles to detect undisplaced cement or implant fractures or minimal loosening, as these findings tend to be subtle. Axial and lateral views were most important, as complications were mostly occult on the AP views. Furthermore, the total extent of the complications was often underestimated, which may limit its effectiveness in monitoring PEEK implants. Baseline comparisons proved especially useful in detecting subtle findings, such as undisplaced cement or implant fractures, which might otherwise be overlooked.

Implant fractures were only detectable on lateral radiograph views in this study due to the presence of gas in the fracture line. In actual clinical settings, these fracture lines would be filled with joint fluid, making them harder to detect radiographically, especially when undisplaced. Implant fractures at the femoral trochlea might be challenging to identify, as shown by the radiographically occult trochlear fracture in our study.

Previous animal studies using PEEK-XLPE knee prostheses have shown that PEEK-on-HXLPE prosthetic materials exhibit low signal intensity and no significant artefacts on MRI scans, enabling clear visualization of surrounding marrow and soft tissues without artefact interference [11]. MRI imaging of all-polymer prostheses might also allow evaluation of the phases of osseointegration implants, from the inflammatory phase, through the proliferative phase, until complete mineralization [12].

The utilization of polymers such as PEEK represents the only currently available implant-based approach to fundamentally avoid clinically relevant metal-related imaging artefacts. Although researchers have attempted to reduce or eliminate these artefacts by adjusting MRI scan data/protocols (metal-artefact reduction scan protocols or MARS) or using titanium, no fundamental solution has been found to eradicate metal artefacts entirely [13, 14]. Metal artefact reduction scan protocols (MARS) typically lead to increased scan duration and diminished resolution. In contrast, standard conventional scan protocols are suitable for all-polymer PEEK implants, enabling rapid and high-resolution imaging [15]. In a metal-free imaging environment, this also enables high-resolution three-dimensional MRI acquisitions, which are typically not feasible around metallic knee prostheses because of severe susceptibility artefacts. Three-dimensional imaging allows isotropic high-resolution assessment with multiplanar reformations, enabling visualization in any orientation and improving analysis of subtle and anatomically complex abnormalities.

Interestingly, our study found that CT and MRI scans still revealed scattered micrometallic artefacts, likely from the metallic surgical instruments leaving particles behind. However, these tiny fragments did not lead to noticeable imaging issues. The minor focal artefacts had no relevant impact on the assessment of structures, as they did not significantly obscure any anatomical details.

PEEK has been widely utilized for intervertebral implants in spine surgery for years, demonstrating reduced artefacts on CT and MRI scanning compared to titanium implants, which enhances the detection of complications such as loosening [16–18]. PEEK materials have also been utilised in other medical fields, including oral oncology, to reduce CT artefacts and enhance image quality, aiding in the radiological interpretation of tissues surrounding implants. This effect can support radiological tumor follow-up and CT-based planning of adjuvant radiotherapy [19].

As demonstrated in our study, CT allows a clear and easy distinction between polyethylene (PE) and PEEK components due to their different radiodensities. In our study, PE demonstrated a mean of -79 HU (range: -109 to -47) and PEEK + 188 HU (range: 142 to 223), due to PEEK’s denser molecular structure (Table 1). Similar findings were reported in a recent post-operative imaging study of an all-PEEK knee prosthesis that showed a mean of 183 HU for PEEK and − 89 HU for the PE bearing [20].

As a limitation, the cadaveric nature of this study imposes several important constraints. First, a cadaveric model cannot simulate bone remodelling or resorption, nor can it mount an inflammatory intra-articular or peri-articular response, which frequently accompanies implant failure in vivo. Consequently, biological processes such as wear-particle–induced osteolysis, synovitis, and bone marrow edema cannot be reproduced. As a result, early biologically driven loosening, characterized by gradual cavitary bone loss and inflammatory soft-tissue changes, is not represented in this model.

Second, the repeated opening and closing of the joint space required to simulate complications inevitably introduced air into peri-articular soft tissues and intra-articular spaces, which would be absent in living patients. Although saline was introduced into the joint and interface spaces, complete elimination of air entrapment was not possible. This artificially introduced air facilitated the detection of certain complications, particularly on radiography and CT, and may overestimate detectability compared with clinical scenarios. In vivo, interface spaces would more typically be filled with synovial fluid, fibrous tissue, or granulation tissue, and the presence of gas would be uncommon.

Third, the methods used to generate cement fractures and interface separation produced greater mechanical displacement than is usually encountered in early clinical failure, where cement fractures are often undisplaced and therefore more difficult to visualize. This may further contribute to an overestimation of how easily cement fractures and early-stage loosening can be detected in clinical settings.

In addition, all imaging assessments were performed by a single experienced fellowship-trained musculoskeletal radiologist, which ensured consistency but precluded assessment of inter-reader variability. Detection performance may therefore differ for general radiologists or in centers without specific experience in polymer-based arthroplasty imaging.

Finally, the sample size was limited to four cadaveric specimens, the simulation methodology has not been externally validated against clinical failure patterns, sequential testing may have introduced cumulative specimen damage, and the use of a novel implant system may limit direct extrapolation to conventional prostheses.

Future studies should therefore focus on evaluating these imaging characteristics in a clinical setting to validate these findings and further explore the efficacy of XR, CT, and MRI for early detection of PEEK implant complications.

In conclusion, all-polymer PEEK knee prostheses may enable high-quality multimodality assessment of implant-related structures and complications in a metal-free implant design context. In a cadaveric setting simulating advanced mechanical failure, this facilitates reliable evaluation using radiography, CT, and MRI.

Conclusion

The all-polymer PEEK knee prosthesis may offer imaging advantages over traditional metal-based knee replacements, particularly in the radiologic detection of complications such as cement fractures and implant loosening. PEEK has imaging characteristics that are radiolucent and demonstrate no relevant artefacts associated with metal implants, allowing for clear visualization of the prosthesis positioning and the prosthesis-cement-bone interface. While radiography detected most complications, MRI and especially CT provided more complete visualization of the full extent of complications. These findings suggest that PEEK implants could enhance post-operative monitoring and early intervention strategies. Further clinical studies are required to validate these results and assess their impact on long-term patient outcomes.

Abbreviations

PEEK

Polyetheretherketone

PE

Polyethylene

TKA

Total Knee Arthroplasty

XR

X-ray / Radiography

CT

Computed Tomography

MRI

Magnetic Resonance Imaging

HU

Hounsfield Unit

Author contributions

All authors contributed to study conception, data acquisition or analysis, manuscript preparation or revision, approved the final manuscript, and agree to be accountable for all aspects of the work.

Funding

This study received financial support from a commercial company. The authors retained full control over study design, data collection, data analysis, interpretation of results, and the decision to submit the manuscript for publication.

Data availability

The imaging datasets generated and analysed during the current study are stored on secure institutional servers. Access to the raw imaging data may be subject to institutional approval and material transfer regulations. Data may be made available from the corresponding author on reasonable request and pending institutional permission.

Declarations

Ethics approval and consent to participate

This cadaveric study was conducted in accordance with applicable national and institutional ethical standards governing the use of human cadaveric material. No living participants or identifiable human data were involved. Therefore, institutional review board (IRB) approval and informed consent were not required.

Consent for publication

Not applicable. This study did not include data from any identifiable person.

Competing interests

CB and DJ received consultancy fees from Invibio Ltd. related to this study.FZ and AB are employees of Invibio Ltd.All other authors declare no additional competing interests.