Aseptic tibial loosening: Radiographic identification remains a diagnostic dilemma

Charles T. Crellin; Jacquelyn S. Pennings; Stephen M. Engstrom; Andrew A. Shinar; Gregory G. Polkowski; J. Ryan Martin

doi:10.1016/j.jorep.2023.100194

View/Download PDF

Buy Reprints

PDF

Translate this page into:

Research Article

2023

:2;

100194

doi:

10.1016/j.jorep.2023.100194

Aseptic tibial loosening: Radiographic identification remains a diagnostic dilemma

Charles T. Crellin, Jacquelyn S. Pennings, Stephen M. Engstrom, Andrew A. Shinar, Gregory G. Polkowski, J. Ryan Martin

Department of Orthopedic Surgery, Vanderbilt University Medical Center, Nashville, TN, 37232, USA

∗Corresponding author: J. Ryan Martin. John.Martin@vumc.org

Received: 2023-5-20, Accepted: 2023-7-8, Published: 2023-7-17

The Author(s)

Disclaimer:
This article was originally published by Reed Elsevier India Pvt. Ltd. and was migrated to Scientific Scholar after the change of Publisher.

Abstract

Aseptic tibial loosening following total knee arthroplasty (TKA) can be challenging to diagnose radiographically. Surgeons often face the clinical scenario of determining implant fixation in patients that do not have serial radiographs and only subtle radiographic features of loosening. Therefore, the purpose of our study was to determine how accurately attendings and trainees are at identifying aseptic tibial loosening.

Two cohorts of patients were identified for this study. Cohort one consisted of twenty patients with aseptic tibial loosening. Cohort two consisted of twenty patients with well-fixed pain-free total knee replacements at one-year follow-up. Surgeons were then given a clinical scenario that all patients are returning with pain at follow-up. The surgeon then reviewed each patient's radiographs and determined whether the tibial component was loose or well-fixed.

Overall, surgeons were significantly more likely to correctly identify well-fixed implants than loose implants (p<0.001). There were no significant differences in the surgeon's vs. trainee's ability to diagnose implant fixation (p = 0.926). However, the percentage of correctly identified loose implants was significantly higher for trainees compared to surgeons (p = 0.023). For surgeons, the proportion of correctly identified well-fixed implants was significantly higher than the proportion of correctly identified loose implants (p<0.001).

Aseptic tibial loosening following primary total knee arthroplasty remains a diagnostic dilemma. Surgeons were significantly less likely to identify implant loosening than trainees. Improving diagnostic accuracy in this patient population is critical to avoid delaying surgery in patients with implant loosening and avoiding unnecessary surgery in patients with well-fixed implants.

Abstract

Highlights

•Aseptic tibial loosening is challenging to diagnose radiographically•Trainees were more likely to identify an implant as loose.•Surgeons were more likely to identify an implant as well-fixed.

Keywords

Aseptic tibial loosening

Primary total knee

Total knee replacement

Total knee arthroplasty

Implant loosening

Show Related Articles from PubMed

1 Introduction

Total knee arthroplasty is one of the most common elective orthopedic surgeries performed every year. It is projected that the number of total knee arthroplasties performed each year will substantially increase. Sloan et al. predict there will be 935,000 primary total knee arthroplasties (TKA) performed annually by 2030.¹ As the rates of primary TKA continue to increase, the revision rates will also increase. For this reason, it is predicted there will be 120,000 revision TKA performed in 2030¹ There are many causes of failure of primary TKA.² Aseptic loosening, in some studies, has been determined to be the leading cause of long-term failure, accounting for 39.9% of revisions.³^,⁴ Aseptic loosening is characterized by a noninfectious failure of fixation leading to pain and debility. This failure of fixation can occur at the implant-cement interface or the cement-bone interface and occurs most commonly at the implant cement interface.^5–7

The diagnosis of aseptic loosening can be very challenging and there is currently no gold-standard preoperative diagnostic test. However, one common method to diagnose aseptic loosening is identification of radiolucenies at the bone-cement interface.^8–13 One source for aseptic loosening is osteolytic bone resorption leading to failure at the cement-bone interface.^14–17 However, this does not explain the more common failure mechanism of implant-cement interface failures which are not typically associated with osteolysis and/or radiolucent lines.¹⁸ Therefore, other diagnostic tests have been explored to identify potential implant loosening. Some surgeons have recommended bone scintigraphy, however this test had demonstrated low diagnostic yield in the setting of aseptic loosening.¹⁹ Other techniques that have been described are radionuclide arthrography, 18- flurorodeoxyglucose positron emission tomography, vibration analysis, radiostereometric analysis, and synovial fluid analysis,^19–24 with variable success. Due to the substantial morbidity associated with delayed diagnosis, identification of aseptic loosening has received lots of attention. Failure to diagnose aseptic loosening can result in unnecessary continued pain until the proper diagnosis is made. Alternatively, incorrectly diagnosed aseptic loosening can result in an unnecessary surgery which places the patient at increased risk of significant complication.

Radiographic identification of aseptic tibial loosening remains a diagnostic dilemma. Therefore, we designed the following study with the primary goal of determining how accurately arthroplasty surgeon are in determining implant fixation in patients presenting with knee pain. Secondarily, we compared increased clinical experience and ability to identify implant fixation. Finally, we evaluated whether there was a trend toward over or underdiagnosis of identification of well-fixed or loose implants.

2 Methods

IRB approval was obtained prior to commencing the following retrospective study. We utilized our institutional joint registry to identify two cohorts of patients with a minimum of 1-year follow up. Cohort one consisted of 20 patients with intraoperatively confirmed aseptic tibial loosening. Cohort two consisted of 20 patients with well-fixed, pain-free total knee arthroplasties returning for routine one-year follow up with no concern for aseptic tibial loosening.

A weight-bearing anteroposterior (AP) and lateral x-ray from each of the 40 patients was de-identified and cropped into a PowerPoint presentation. The images were randomly inserted into the file. All patients underwent primary and or revision TKA between 2016 and 2021. For the aseptic loosening cohort, the radiographic series (AP and lateral) immediately preceding revision surgery was selected. This imaging series was considered to best represent implant loosening and was confirmed on the preoperative notes to be consistent with implant loosening. Additionally, we chose to utilize a single image series rather than sequential x-rays. This was felt to better represent a new patient clinical workup in which patients often present with no preceding x-rays and surgeons must direct clinical care on a single image series.

This PowerPoint was then presented to 20 participants (10 attending surgeons and 10 trainees). The trainees were comprised of upper-level residents going into adult reconstruction (N = 5) as well as adult reconstruction fellows (N = 5). The test was administered in the following format. Each surgeon was given the same clinical scenario for all patients (well-fixed and loose), “A patient comes to you for a second opinion. The patient has anterior knee pain and on exam the knee feels stable and well-balanced. The patient is unable to recall the previous surgeon's name and as a result you are unable to get any previous radiographs of the knee.” The participant was then shown one x-ray series at a time and asked to determine whether the tibial implant was loose or well-fixed prior to moving to the next x-ray. Their answers were recorded on an excel spreadsheet. This procedure was performed both in person and on zoom based on the location of the test-taker. Expansion of this study outside of our institution was proposed as a method to limit the bias of a single institution study by allowing multiple participants from varied centers throughout the US to participate. We specifically included attending surgeons from outside institutions with research interests in implant loosening.

2.1

2.1 Analysis

The original classification of loose or well-fixed was recorded for each participant response during administration of the PowerPoint test. For each participant, responses were then scored as correct if they correctly identified a loose implant as loose or a well-fixed implant as not loose. The area under the curve (AUC), 95% confidence interval (CI), p-value, Youden's index, sensitivity, specificity, positive and negative likelihood ratios (+LR and -LR), and positive and negative predicted values (PPV and NPV) were calculated for the full sample, as well as for the attending surgeons and trainee surgeons separately. These values were also computed for each individual participant to obtain the ranges of these values for each participant group (attendings vs trainees). The AUC for the pooled group of attendings was compared to the AUC for the pooled group of trainees using the DeLong et al. (1988) method to evaluate if one group was more accurate than the other.²⁵

To evaluate whether there was a trend toward over or underdiagnosis of loose implants, the mean proportion of correctly identified loose implants was compared to the mean proportion of correctly identified well-fixed implants using an independent samples t-test. The sample was then divided into attendings vs. trainees and the same t-test was performed on each group in order to determine if attendings and trainees were more accurate in determining loose or well-fixed implants. Finally, the percentage of correct responses was computed for each of the 40 patient radiograph series (overall and by surgeons vs trainees). The patient radiographs with the highest and lowest percent of correct classifications (loose or well-fixed) were summarized and visually inspected in order to identify subtle radiographic features that may help explain why they were not correctly identified. Alpha was set at 0.05. MedCalc version 18.2.1 and IBM SPSS version 28 were used for statistical analysis.

3 Results

Overall, 20 there were 20 participants in the study. Of these, 10 were surgeons and 10 were trainees. Each participant rated all 40 radiographic series, which were made up of 20 loose and 20 well-fixed implants. The overall accuracy of the entire cohort (20 surgeons) was 67% (AUC = 0.67, 95% CI = 0.62-0.70, p<0.001). When evaluating the surgeons, their overall accuracy was also 67% (95% CI = 0.62-0.71, p<0.001) and the trainees accuracy was also 67% (95% CI = 0.62-0.72, p<0.001). The accuracy of the individual attendings ranged from 53% to 85% while the trainee's accuracy ranged from 53% to 80%. The sensitivity, specificity, +LR, -LR, PPV, and NPV are shown in Table 1. Overall, there was not a significant difference between the accuracy of surgeons and trainees in determining loose versus well-fixed implants, (p = 0.926).

Table 1 AUC, sensitivity, specificity, likelihood ratios and predicted values overall and by training level.

	Overall	Surgeon	Trainee
AUC	0.67	0.67	0.67
SE	0.02	0.02	0.02
95% CI	0.64 to 0.70	0.62 to 0.71	0.62 to 0.72
p-value	<0.001	<0.001	<0.001
Youden's index	0.34	0.34	0.34
Sensitivity	57%	51%	63%
Specificity	77%	83%	71%
+LR	2.47	2.97	2.17
-LR	0.56	0.60	0.52
PPV	71.2	74.8	68.5
NPV	64	62.6	65.7

Across all participants, the mean proportion of correctly identified well-fixed implants (M = 0.77, SD = 0.19) was significantly greater than the mean proportion of correctly identified loose implants (M = 0.57, SD = 0.22), p = 0.004. The difference was even more pronounced when evaluating surgeons with the mean proportion of correctly identified well-fixed implants (M = 0.83, SD = 0.15) was significantly greater than the mean proportion of correctly identified loose implants (M = 0.51, SD = 0.23), p<0.001. However, for trainees there was not a significant difference between the proportion of correctly identified well-fixed (M = 0.71, SD = 0.25) compared to loose (M = 0.63, SD = 0.26) implants, p = 0.332.

When evaluating the responses of all 20 participants, the percentage of correctly identified well-fixed implants ranged from 35% to 100% and the percentage of correctly identified loose implants ranged from 10% to 95%. The following two image series were noted to be incorrectly identified outliers. Figs. 1 and 2 represent one of the well-fixed implants which was incorrectly identified as loose 12 out of 20 times (Figs. 1 and 2.) Figs. 3 and 4 represent one of the loose implants which was incorrectly identified as well-fixed 18 out of 20 times (Figs. 3 and 4).

The following AP (a.) left knee radiographs represent 1-year follow-up images demonstrating a well-fixed left TKA.

The lateral (b.) left knee radiographs represent 1-year follow-up images demonstrating a well-fixed left TKA.

The following AP (a.) left knee radiographs represent aseptic tibial loosening. The lateral view shows some anterior heterotopic bone formation anterior to the tibial tray which was noted in most of the aseptic loosening cases.

The following lateral (b.) left knee radiographs represent aseptic tibial loosening. The lateral view shows some anterior heterotopic bone formation anterior to the tibial tray which was noted in most of the aseptic loosening cases.

4 Discussion

Radiographic identification of aseptic loosening remains a diagnostic challenge.²²^,²⁶Radiographs often fail to demonstrate radiolucent lines consistent with failure. Furthermore, radiolucent lines may be lacking in patients that fail at the implant-cement interface. Our study confirms that the overall accuracy of surgeon identification of implant fixation status was 67%. Therefore, nearly one-third of patients were incorrectly identified in our sample. Furthermore, the accuracy of arthroplasty trained surgeons was no better than the trainee's ability to identify fixation status.

One potential limitation that was frequently noted when administering the PowerPoint to participants was that they wanted serial radiographs to compare with. However, we specifically designed the following study to replicate the common clinical scenario without serial radiographs. Most patients present without and preoperative or postoperative radiographs available for comparison. Furthermore, we have noted the lack of progressive radiolucent lines in patients with aseptic tibial failure at the implant-cement interface. However, there are subtle findings that have recently helped us in the diagnosis of aseptic loosening including heterotopic bone formation anterior and posterior to the tibial tray and medial and lateral cyst formation. We believe that utilization of some of these newer radiographic findings may aide in the diagnosis of aseptic loosening in certain clinical scenarios.

We note that in our sample participants were 20% more likely to correctly identify well-fixed implants compared to loose implants (77.0% vs 56.8%). When this finding was evaluated by subgroup of surgeons vs. trainee, it was found that surgeons were much more likely to correctly identify well-fixed compared to loose implants but that there was no difference in the proportion of correctly identified well-fixed and loose implants for trainees. It is possible and likely that surgeons err on misdiagnosing a loose implant more frequently for two potential reasons. First, as discussed previously, radiographic features of radiolucencies are often absent in patients that fail at the implant-cement interface. Furthermore, we specifically did not include any failures with accompanying varus collapse as these are substantially easier to radiographically identify. Secondly, taking a patient to the operating room for implant loosening when a patient has a well-fixed implant is a potentially morbid mistake that should be avoided and therefore, we are more likely to avoid this potential mistake.

Currently, there is not a gold-standard test for diagnosing aseptic loosening. We have identified previously that in contemporary TKA, the implant-cement interface is the weak-link of fixation. Furthermore, obvious radiographic features of implant migration, progressive radiolucencies, and cement mantle fractures are often absent. We do not have a test to date that can accurately and reliably detect implant loosening. Therefore, surgeons are forced to interpret radiographs with inherent diagnostic limitations. We believe that some of the subtle radiographic features discussed above may help in the diagnosis and we plan on potentially expanding this project in the future to presenting these radiographic findings and then repeating the testing procedure.

4.1

4.1 Limitations

While we believe the following study provides unique insight into the diagnostic challenges of radiographic identification of aseptic tibial loosening, there were a few limitations to this study. First, we included only two radiographic views to determine implant fixation. While we understand that this is a limitation for identifying progressive radiolucent lines, we specifically chose cases of implant-cement interface failures which were not accompanied by these radiographic features. Furthermore, the common clinical scenario we encounter daily are patients presenting with a single radiographic series and a painful TKA. Therefore, we do not believe this to be a true limitation of the study. Another potential limitation was the lack of fluoroscopic guided X-rays. Our institution, as well as many others, does not have this technology. Therefore, radiographs are often not perfectly aligned to evaluate the interfaces. It is possible that obtaining perfect AP and lateral radiographs may have increased the diagnostic accuracy amongst surgeons and trainees. We believe that these radiographs are routinely obtained and are representative of current clinical scenarios. Finally, some participants associated specific implants with increased or decreased implant fixation.⁷^,²⁷ Therefore, these surgeons may have been inherently biased during the test.

5 Conclusion

Our results indicate that identification of aseptic tibial loosening following primary total knee arthroplasty remains a diagnostic dilemma. Radiographic features of progressive radiolucent lines may be absent in the setting of implant-cement interface failures or in patients that present with a single radiographic series. There are subtle radiographic features that appear to be present in cases of aseptic tibial failure, including anterior heterotopic bone formation and medial and lateral cyst formation that might indicate implant loosening. Long-term, a gold-standard test for identifying implant loosening is necessary to improve diagnostic accuracy and limit patient morbidity by avoiding falsely identifying implant loosening or fixation.

Author statement

Charles T. Crellin, MD: Conceptualization, investigation, writing. Jacquelyn S. Pennings, PhD: Formal analysis Stephen M. Engstrom, MD: Supervision Andrew A. Shinar, MD: Supervision Gregory G. Polkowski, MD: Supervision J. Ryan Martin, MD: Conceptualization, investigation, writing.

Funding statement

No funding was received for this study.

Funding/sponsorship

oThis research did not receive any specific grant from funding agencies in the public, commercial or not-for-profit sectors-Informed consent (Patient/Guardian), mandatory only for case reports/clinical imagesoNot required-Institutional Ethical Committee Approval (for all human studies)oApproved

References

Sloan M., Premkumar A., Sheth N.P., . Projected volume of primary total joint arthroplasty in the U.S., 2014 to 2030. J Bone Joint Surg Am. 2018;100(17):1455-1460.
[Google Scholar]
Bozic K.J., et al . The epidemiology of revision total knee arthroplasty in the United States. Clin Orthop Relat Res. 2010;468(1):45-51.
[Google Scholar]
Hampton C.B., et al . Aseptic loosening at the tibia in total knee arthroplasty: a function of cement mantle quality? J Arthroplasty. 2020;35(6S):S190-S196.
[Google Scholar]
Sharkey P.F., et al . Why are total knee arthroplasties failing today--has anything changed after 10 years? J Arthroplasty. 2014;29(9):1774-1778.
[Google Scholar]
Cox Z.C., et al . Varus collapse in total knee arthroplasty: does fixation or bone fail first? J Arthroplasty. 2022;37(1):162-167.
[Google Scholar]
Guha A.R., Debnath U.K., Graham N.M., . Radiolucent lines below the tibial component of a total knee replacement (TKR)--a comparison between single-and two-stage cementation techniques. Int Orthop. 2008;32(4):453-457.
[Google Scholar]
Hoskins W., et al . High incidence of radiolucent lines at the implant-cement interface of a new total knee replacement. ANZ J Surg. 2020;90(7-8):1299-1302.
[Google Scholar]
Bach C.M., et al . Reproducible assessment of radiolucent lines in total knee arthroplasty. Clin Orthop Relat Res (434):183-188.
[Google Scholar]
Chalmers B.P., et al . A novel percentage-based system for determining aseptic loosening of total knee arthroplasty tibial components. J Arthroplasty. 2017;32(7):2274-2278.
[Google Scholar]
Ecker M.L., et al . Long-term results after total condylar knee arthroplasty. Significance of radiolucent lines. Clin Orthop Relat Res (216):151-158.
[Google Scholar]
Ewald F.C., . The Knee Society total knee arthroplasty roentgenographic evaluation and scoring system. Clin Orthop Relat Res (248):9-12.
[Google Scholar]
Hunter J.C., et al . Loosening of the total knee arthroplasty: detection by radionuclide bone scanning. AJR Am J Roentgenol. 1980;135(1):131-136.
[Google Scholar]
Meneghini R.M., et al . Development of a modern knee society radiographic evaluation system and methodology for total knee arthroplasty. J Arthroplasty. 2015;30(12):2311-2314.
[Google Scholar]
Cyndari K., . Mechanisms of Aseptic Loosening in Total Knee Arthroplasty. 2017
[Google Scholar]
Dyskova T., et al . Inflammation time-axis in aseptic loosening of total knee arthroplasty: a preliminary study. PLoS One. 2019;14(8)
[Google Scholar]
Goodheart J.R., Miller M.A., Mann K.A., . In vivo loss of cement-bone interlock reduces fixation strength in total knee arthroplasties. J Orthop Res. 2014;32(8):1052-1060.
[Google Scholar]
Kienzle A., et al . High rates of aseptic loosening after revision total knee arthroplasty for periprosthetic joint infection. JB JS Open Access. 2020;5(3)
[Google Scholar]
Kopinski J.E., et al . Failure at the tibial cement-implant interface with the use of high-viscosity cement in total knee arthroplasty. J Arthroplasty. 2016;31(11):2579-2582.
[Google Scholar]
Arami A., et al . Knee implant loosening detection: a vibration analysis investigation. Ann Biomed Eng. 2018;46(1):97-107.
[Google Scholar]
Chalmers P.N., et al . Evaluation of the role for synovial aspiration in the diagnosis of aseptic loosening after total knee arthroplasty. J Bone Joint Surg Am. 2015;97(19):1597-1603.
[Google Scholar]
Pijls B.G., Plevier J.W.M., Nelissen R., . RSA migration of total knee replacements. Acta Orthop. 2018;89(3):320-328.
[Google Scholar]
Sterner T., et al . The role of [18F]fluoride positron emission tomography in the early detection of aseptic loosening of total knee arthroplasty. Int J Surg. 2007;5(2):99-104.
[Google Scholar]
Ryd L., et al . Roentgen stereophotogrammetric analysis as a predictor of mechanical loosening of knee prostheses. J Bone Joint Surg Br. 1995;77(3):377-383.
[Google Scholar]
Fontalis A., Haddad F.S., . Roentgen stereophotogrammetric analysis: still a very valuable tool in the orthopaedic research armamentarium. Bone Joint Res. 2022;11(4):210-213.
[Google Scholar]
DeLong E.R., DeLong D.M., Clarke-Pearson D.L., . Comparing the areas under two or more correlated receiver operating characteristic curves: a nonparametric approach. Biometrics. 1988;44(3):837-845.
[Google Scholar]
Rowlands A., Duck F.A., Cunningham J.L., . Bone vibration measurement using ultrasound: application to detection of hip prosthesis loosening. Med Eng Phys. 2008;30(3):278-284.
[Google Scholar]
Martin J.R., et al . Chitranjan S. Ranawat award: motion during total knee cementing significantly decreases tibial implant fixation strength. J Arthroplasty. 2022;37(6S):S12-S18.
[Google Scholar]