Translate this page into:
An evaluation of the diagnostic performance of the triphasic bone scintigraphy in patients suspected of aseptic total knee arthroplasty loosening
⁎Corresponding author: Leendert Blankevoort. l.blankevoort@amsterdamumc.nl
-
Received: ,
Accepted: ,
This article was originally published by Reed Elsevier India Pvt. Ltd. and was migrated to Scientific Scholar after the change of Publisher.
Abstract
Abstract
Total Knee Arthroplasty (TKA) is a widely conducted and successful orthopedic procedure. However, aseptic loosening, a common cause of TKA failure, necessitates revision surgery. Diagnostic accuracy of triphasic bone scintigraphy, a common imaging modality for aseptic loosening detection, remains controversial. This study investigated the diagnostic accuracy of bone scintigraphy when separately evaluated by a nuclear physicist and an orthopedic surgeon, and the interrater reliability between the two.
Patients undergoing knee revision surgery due to suspected aseptic loosening at three medical centers from 2006 to 2023 were included. Relevant demographic, clinical, and procedural data were extracted from the records. The bone scintigraphy results as noted by the nuclear physicist and orthopedic surgeon were used as index test and intraoperative findings of loosening were used as reference tests. Accuracy, sensitivity, specificity, positive predictive value, and negative predictive value were calculated, and kappa's agreement was assessed.
Out of 611 revision TKAs, 59 cases were analyzed. The nuclear physicist's evaluation of bone scintigraphy had a sensitivity of 73 %, specificity of 0 %, positive predictive value of 93 %, negative predictive value of 0 %, and diagnostic accuracy of 69 %. The orthopedic surgeon's evaluation showed higher sensitivity, specificity, positive and negative predictive values, and an accuracy of 84 %. Agreement levels were moderate (kappa = 0.46) between the nuclear physicist's and orthopedic surgeons evaluation.
The diagnostic accuracy of bone scintigraphy for aseptic loosening is 84 % when evaluated by the orthopedic surgeon compared to 69 % for the nuclear physicist's evaluation. Kappa's agreement between the two was moderate.
Keywords
Bone scintigraphy
Accuracy
Total knee arthroplasty
Aseptic loosening
1 introduction
Total Knee Arthroplasty (TKA) is one of the most frequent and successful orthopedic procedures currently being conducted.1 Several published studies and registries report and project a rising incidence of knee arthroplasty.2–4 Although successful in most patients, persistent pain after arthroplasty is a common complication that affects up to 27 % of TKA patients, with reported revision rates up to 13 % within 10 years.5–8
Aseptic loosening is reported as one of the most frequent modes of TKA failure, with reported incidences of 20.4 % due to loosening of the tibial component and 8.8 % due to loosening of the femur component.9 Aseptic implant loosening often requires major revision surgery and places a significant burden on patients and healthcare systems worldwide.10,11 Therefore, accurate and timely detection is crucial in guiding management decisions in patients with persistent post-arthroplasty pain.
A triphasic bone scintigraphy is a commonly used nuclear imaging modality for aiding the diagnosis of aseptic loosening of knee implants.12 A bone scintigraphy is evaluated for radioisotope uptake around the prosthetic component indicating focally increased bone activity in this area. Increased uptake might indicate component loosening but also infection or healing of micro fractures.13 Bone scintigraphy is widely available and relatively inexpensive, making it an attractive modality for orthopedic surgeons to aid the diagnosis of aseptic loosening. However, with conflicting results reported in literature, its diagnostic accuracy as an imaging modality on its own is still a matter of debate.22
Available literature usually reports solely the judgment of component loosening by the nuclear physicist in a blinded research setting. Yet in daily practice, the decision to proceed to revision surgery is made by the orthopedic surgeon who is often evaluating the judgment of the nuclear physicist, images, and levels of tracer activity themselves. Therefore, the judgment of implant loosening of the orthopedic surgeon should be considered as well when reporting on the diagnostic accuracy of the bone scintigraphy in clinical practice. Moreover, the agreement between the nuclear physicist and the orthopedic surgeon should be reported as this may be of clinically relevant as diagnostic accuracy results may be similar while agreement on individual cases can be poor.
Therefore, this study aimed to answer two research questions: (1) what is the diagnostic accuracy of the bone scintigraphy in a daily practice when evaluated separately by the nuclear physicist and the orthopedic surgeon? and, (2) what is the interrater variability between the nuclear physicist and the orthopedic surgeon?
The hypothesis is that diagnostic accuracy of the bone scintigraphy is sufficient to aid the diagnosis of aseptic loosening as its diagnostic accuracy is expected to be > 0.5. Based on experience of the co-authors, the interrater reliability between nuclear physicist and orthopedic surgeon is hypothesized to be fair to moderate.
2 methods
ethical statement
The institutional review board of the Amsterdam University Medical Centers (AUMC) approved this retrospective study (W22_231) and waived the need for informed consent due to the retrospective nature of this study. This study was performed in accordance with the recommendations for Conduct, Reporting, Editing and Publication of Scholarly Work in Medical Journals and the European General Data Protection Regulation (GDPR).14,15
2.1 patient screening and inclusion
The data from all patients receiving knee revision surgery for aseptic loosening at the Amsterdam University Medical Center, Academisch Medisch Centrum and the Vrije Universiteit Medisch Centrum between 2006 and 2023 were extracted from a national registry for orthopedic interventions (Nederlandse Registratie voor Orthopedische Interventies [LROI]) and screened for eligibility.16 Patients were eligible and therefore included within this study if the bone scintigraphy was performed at least twelve months after previous knee arthroplasty surgery, to reduce the chance of false positives due to positive uptake after surgery.17 Also, bone scintigraphy needed to be performed at least twelve months prior to revision surgery. Patients with a positive synovial fluid culture after preoperative arthrocentesis were excluded because these patients were considered to be septic.
If a single patient had multiple bone scintigraphy's and revision surgeries within the inclusion period, the subsequent scintigraphy and revision surgery were grouped as one case and the two were included as a separate case in the study.
2.2 Data collection
All records of included patients were screened and demographic data (gender, year of birth), latest preoperative weight and height, smoking status, and date of latest joint specific arthroplasty surgery were extracted.
For the index test, (three phase bone scintigraphy) dosage and type of used radiopharmaceutical, date of execution of bone scintigraphy and the assessment of loosening of all components (separate and overall) by both the nuclear physicist and (if available) by the orthopedic surgeon were extracted. If applicable, the use of any other nuclear modalities (e.g. (Single Photon Emission Computed Tomography) scan with a CT (Computed Tomography) scan [SPECT/CT]) were registered. In relation to the reference test, date of revision surgery, intraoperative assessment of loosening of all components (separate and overall), and the results of perioperative cultures were extracted.
2.3 index test
In this study, the results of a three-phase bone scintigraphy (and if applicable, combined with SPECT/CT) as reported in a radiology report (RR) by the nuclear physicist and the preoperative orthopedic surgeon's prediction (SP) were used as index test.
RR was defined as positive for prosthesis component loosening if reported so for any component by the nuclear physicist. The diagnosis was based on the comprehensive evaluation of all images by the academic nuclear physicist, observing increased tracer activity around the prothesis in both the early and later phases. The bone scintigraphy's were evaluated and reported by nuclear physicists who were not blinded to previous imaging, clinical history, pattern of complaints and physical examination.
SP was defined as positive for prosthesis component loosening if reported so for any component based in the notes of the orthopedic surgeon following the bone scintigraphy. Although notes were screened for SP based on solely the evaluation of the bone scintigraphy, orthopedic surgeons were not blinded for RR nor for previous imaging, clinical history, pattern of complaints and physical examination. SP was registered as missing if no separate evaluation of the bone scintigraphy was registered in the notes by the orthopedic surgeon.
The bone scintigraphy scans were evaluated by one of the academic nuclear physicists (RR), and orthopedic staff members (SP).
2.3.1 Imaging protocol
2.3.1.1 Three phase bone scintigraphy
A dual-headed gamma camera (Symbia Intevo, Siemens Healthcare GmbH, Munich, Germany), equipped with a low energy high-resolution (LEHR) or low energy all-purpose (LEAP) collimator, was used. Patients were intravenously injected with 99mTc-oxidronate ([99Tc-m] HDP), with dosages dependent on body weight, ranging from 60 MBq to 900 MBq. Approximately 2 min after administration of the [99Tc-m] HDP), the first-phase dynamic and second-phase static images were taken. After approximately 3 h, the third-phase static scan was made. Between the dynamic phase and second-phase static scan, patients were encouraged to walk, drink at least a liter (33.8 oz) of water and use the toilet. All three phases were performed in a supine, feet-first orientation. In each phase, standard anterior, posterior, and lateral images were taken using a 15 % window at a 140 keV photopeak (Tc99m-NMG) with an image matrix size of 128x128 pixels for the dynamic phase and 256x256 pixels for the static phases.
2.3.1.2 SPECT/CT
If deemed necessary by the nuclear physicist, SPECT/CT images were acquired using a hybrid SPECT/CT system (Siemens Symbia, Munich, Germany) equipped with a LEHR collimator. The CT parameters used were 30 mA, 130 keV, 512 × 512 matrix size, and 1 mm slice thickness. SPECT/CT was performed with a matrix size of 128 × 128, 1.0 zoom, 40 s per frame, and 120 frames at 3° intervals. Image reconstruction was performed using vendor-recommended iterative reconstruction algorithms with attenuation correction applied. No post-reconstruction filter was used.
2.4 reference test
Intraoperative findings (IF) of loosening by the orthopedic surgeon performing the revision TKA (rTKA) was used as the only reference test. A component was considered loose when it was reported as such by the orthopedic surgeon in the surgical report. If there was no statement of evidence of loosening of any or both components within the surgical report, then the concerning components were scored as not loose. If an unclear statement was made regarding loosening, the applicable component was scored as not loose. Orthopedic surgeons were not blinded for the RR and/or SP.
2.5 statistical analysis
Post-hoc sample size evaluations were performed for both with and without the use of SPECT/CT using ClinCalc Post-hoc Power Calculator (Alpha: 0.05), resulted in an estimated post-hoc power of 100 % and 6.5 %, respectively.18
Frequencies of true positives (TP), false positives (FP), true negatives (TN) and true positive (TP) findings by both the nuclear physicist and orthopedic surgeon were determined. Accuracy, sensitivity, specificity, positive predictive value (PPV) negative predictive value (NPV) were calculated.
Inter-observer agreement was calculated using Kappa's between the verdict of component loosening in the RR and SP. Due to insufficient post hoc power, no further logistic regression modeling was employed to assess the potential impact of SPECT/CT on the accuracy of the RR or SP.
Parametric data were presented as mean with standard deviation (SD) and non-parametric as median with interquartile range (IQR). Categorical data were presented as frequencies with proportions. A p-value <0.05 was considered statistically significant. Data were analyzed with R for Windows, version 4.2.3, using the and “irr”, “dplyr”, “glm” packages.19
3 results
3.1 patient screening and inclusion
In the three hospitals, 611 rTKAs were performed, of which 175 were for suspected aseptic loosening. Among these, in 79 knee cases, an adjacent bone scintigraphy was deemed necessary by the treating orthopedic surgeon and reported in the patient records. After assessment of eligibility against inclusion and exclusion criteria, 59 knee cases were included for analysis (Fig. 1).
3.2 characteristics
The mean age at the latest knee arthroplasty surgery prior to the revision surgery was 59.9 years (SD 9.1). The mean age at the time of the bone scan and revision TKA were 65.5 years (SD 9.2) and 66.0 years (SD 9.1), respectively. For all included cases, there was no pre-operative suspicion of infection. In 9 cases, a two-stage revision TKA was performed due to intra-operatively suspected infection, yet in all cases, intra-operatively taken cultures were proven negative. The median dosage of [99Tc-m] HDP was 524 MBq (IQR: 503–573). In 18 cases (30 %), an additional SPECT/CT was performed (Table 1).
| Patient characteristics | Overall (n = 59), n (%) |
| Age at latest knee arthroplasty surgery, mean (SD) | 59.9 (9) |
| Age at revision surgery after bone scintigraphy, mean (SD) | 66 (9) |
| Age at bone scintigraphy, mean (SD) | 65.5 (9) |
| Time interval between bone scintigraphy and revision surgery in months, median (IQR) | 4.0 (2–4) |
| Gender | |
| Female | 30 (51) |
| Male | 29 (49) |
| Smoking status | |
| Yes | 11 (13) |
| No | 48 (87) |
| Body Mass Index, median (IQR) | 28.0 (27–34) |
| Laterality | |
| Right | 27 (46) |
| Left | 32 (54) |
3.3 accuracy
3.3.1 Radiology rapport
Analyzing 59 cases of suspected TKA loosening, comparing RR and IF yielded: 41 TP, 0 TN, 3 FP, and 15 FN. These findings indicate a sensitivity of 73 %, specificity of 0 %, PPV of 93 %, NPV of 0 %, and diagnostic accuracy of 69 % (Table 2a).
| Intra-operative findings (IF) | |||
| Fixed | Loose | ||
| Radiology Report (RR) | Fixed | 0 | 15 |
| Loose | 3 | 41 | |
3.3.2 Surgeons’ prediction
SP was not registered in 13 cases, so these were therefore excluded from the accuracy analysis. Consequently, evaluating 46 cases of suspected TKA loosening, comparing SP and IF revealed: 38 TP, 1 TN, 2 FP, and 5 FN. This yielded sensitivity of 88 %, specificity of 33 %, PPV of 95 %, NPV of 17 %, and accuracy of 85 % (Table 2b).
| Intra-operative findings (IF) | |||
| Fixed | Loose | ||
| Prediction by Orthopedic Surgeon (SP) | Fixed | 1 | 5 |
| Loose | 2 | 38 | |
3.4 agreement
3.4.1 Overall agreement
The determination of agreement levels for component loosening between RR and SP resulted in a kappa statistic of 0.46 (p = 0.0006; n = 46), denoting a moderate level of agreement. Conversely, the kappa value between RR and IF was −0.09 (p = 0.3). An assessment of agreement between SP and IF yielded a kappa value of 0.15 (p = 0.28).
3.4.2 Per component
Among the cases examined, the RR suspected no loosening in 15 cases (25 %), while the IF reported no loose components in only 2 cases (3 %). Specifically, the tibial component was predicted as loose in 27 cases (46 %), with the IF reporting loosening in 37 cases (63 %). On the other hand, the femur component was predicted as loose in only 1 case (2 %) but was reported as loose in 4 cases (7 %) by the IF. Both the femur and tibia components were predicted as loose in 16 cases (27 %), and the IF reported their loosening in 16 cases (27 %) (Table 3). The analysis of agreement levels resulted in a kappa statistic between the RR and IF for the loose tibia component of 0.04 (p = 0.72). In contrast, the kappa statistic for the loose femur component was 0.49 (p = 0.0002).
| Component Diagnosed as Loose | Radiologist Report (RR) (n = 59) | Intra-operative Findings (IF) (n = 59) |
| None | 15 (25.4 %) | 2 (3.4 %) |
| Tibia | 27 (45.8 %) | 37 (62.7 %) |
| Femur | 1 (1.7 %) | 4 (6.8 %) |
| Tibia/femur | 16 (27.1 %) | 16 (27.1 %) |
4 discussion
This pragmatic study evaluated the diagnostic accuracy of bone scintigraphy when used in clinical practice. It reveals that bone scintigraphy has a diagnostic accuracy of 84 % in detecting aseptic loosening when evaluated by an orthopedic surgeon, compared to 69 % when evaluated by a nuclear physicist. There is moderate agreement (Kappa = 0.46; p = 0.0006) between the orthopedic surgeons and nuclear physicists’ preoperative predictions of TKA component looseness. These findings support our hypothesis that the diagnostic accuracy of bone scintigraphy is >50 %, thus adding value to the diagnostic pathway of knee implant loosening. The results also underscore the hypothesized fair to moderate agreement between the preoperative assessments of the orthopedic surgeon and the nuclear physicist.
These results suggest better sensitivity but poorer specificity when the nuclear physicist interpreted the bone scan, compared to the orthopedic surgeon. This implies that bone scintigraphy, as interpreted by the nuclear physicist, is more likely to identify patients who truly have aseptic loosening but also incorrectly labels some patients without aseptic loosening as loose (false positives). However, such false positives were less frequent with the surgeon's interpretation. Our assumption is that the orthopedic surgeons slightly superior accuracy arises from their ability to access a more extensive range of clinical details. This assumption is supported by the reported moderate level of agreement between the two.
Limitations of our study include the lack of blinding for the surgeons regarding the results of the bone scintigraphy as noted by the nuclear physicist, which may have introduced bias in their intraoperative assessments of implant loosening. It is also likely that the orthopedic surgeons, as the primary responsible care providers, were more aware and therefore more inclined to consider the results of clinical and other radiological tests in their diagnostic decision-making. Additionally, potential variability in the surgeons' experience and technique might have influenced the intraoperative findings of what is determined as loose or fixed, potentially affecting the accuracy and agreement results.20 Furthermore, the impact of additional use of SPECT/CT could not be evaluated in the current study due to insufficient power.
The findings of this study align with the results of previous systematic reviews. Barnsley et al. conducted a comprehensive review of nuclear imaging modalities for evaluating aseptic TKA loosening, reporting sensitivity and specificity rates for bone scintigraphy between 0.76 and 1.00 and 0.33 and 1.00, respectively.12 While our study did not confirm these findings, Barnsley et al. found that SPECT/CT provided improved accuracy over bone scintigraphy. This conclusion is further supported by Anzola et al.'s recent systematic review and meta-analysis, which assessed SPECT/CT's effectiveness in diagnosing knee conditions, including loosening, in non-infected knees. They reported a pooled sensitivity of 0.86 and specificity of 0.90 across eight studies for SPECT/CT.21 A more recent diagnostic test accuracy review and meta-analysis by Buijs et al. expands on these findings by incorporating both nuclear and non-nuclear imaging modalities, highlighting that MRI and SPECT/CT are currently the most accurate diagnostic tools for assessing aseptic loosening in knee arthroplasty, despite the overall low certainty of evidence due to high risk of bias and heterogeneity among studies.22
While the outcomes derived from both bone scintigraphy and SPECT/CT appear commendable, less cumbersome and cost-effective alternatives, such as quantified and visualized assessment of induced implant motion, have been proposed as potential aids in the diagnosis of implant loosening.13,23–27
Overall, the results of this pragmatic study provide insights into the utility of bone scintigraphy in diagnosing aseptic loosening post-TKA when evaluated in a clinical routine, but also highlight the importance of research and development of additional diagnostic methods.
5 conclusions
The primary results from our study show that bone scintigraphy, whether evaluated by a nuclear physicist or an orthopedic surgeon, demonstrates adequate sensitivity but has limited specificity and negative predictive value. This indicates that bone scintigraphy may be useful for ruling in aseptic loosening but is unreliable for ruling out the condition.
Funding
This study was funded by an internal TKI-PPP grant from the Amsterdam University Medical Center. Allowance made available by Health-Holland, Top Sector Life Sciences & Health, to stimulate public-private partnerships.
conflict of interest
Authors Leendert Blankevoort, Arthur Kievit and Matthias Schafroth are listed as inventors on a patent for a loading device that can be used to quantify and visualize implant displacement.
data availability statement
The data that support the findings of this study are available on request from the corresponding author.
Ethical statement
The institutional review board of the Amsterdam University Medical Centers (AUMC) approved this retrospective study (W22_231) and waived the need for informed consent due to the retrospective nature of this study. This study was performed in accordance with the recommendations for Conduct, Reporting, Editing and Publication of Scholarly Work in Medical Journals and the European General Data Protection Regulation (GDPR).
CRediT authorship contribution statement
George S. Buijs: Conceptualization, Methodology, Formal analysis, Writing – original draft. Arthur J. Kievit: Conceptualization, Validation, Resources, Supervision, Writing – review & editing. Matthias U. Schafroth: Conceptualization, Validation, Resources, Supervision, Writing – review & editing. Leendert Blankevoort: Conceptualization, Validation, Resources, Project administration, Supervision, Writing – review & editing.
References
- Future projections of total hip and knee arthroplasty in the UK: results from the UK Clinical Practice Research Datalink. Osteoarthritis Cartilage. 2015/04/01/2015;23(4):594-600.
- [Google Scholar]
- Projections of primary and revision hip and knee arthroplasty in the United States from 2005 to 2030. JBJS. 2007;89(4)
- [Google Scholar]
- Trends in total knee arthroplasty implant utilization. J Arthroplasty. 2015/05/01/2015;30(5):739-742.
- [Google Scholar]
- Revision rates after total joint replacement: cumulative results from worldwide joint register datasets. Article. J Bone Jt Surg. 2011;93 B(3):293-297.
- [Google Scholar]
- Revision surgery after total joint arthroplasty: a complication-based analysis using worldwide arthroplasty registers. J Arthroplasty. 2013/09/01/2013;28(8):1329-1332.
- [Google Scholar]
- Persistent pain after joint replacement: prevalence, sensory qualities, and postoperative determinants. Pain. 2011;152(3)
- [Google Scholar]
- Painful prosthesis: approaching the patient with persistent pain following total hip and knee arthroplasty. Clin Cases Miner Bone Metab. May 2013;10(2):97-110.
- [Google Scholar]
- Population-based 10-year cumulative revision risks after hip and knee arthroplasty for osteoarthritis to inform patients in clinical practice: a competing risk analysis from the Dutch Arthroplasty Register. Acta Orthop. Jun 2021;92(3):280-284.
- [Google Scholar]
- Financial analysis of revision knee surgery based on NHS tariffs and hospital costs Does it pay to provide a revision service? Article. Bone Joint Lett J. 2015;97-B(2):197-201.
- [Google Scholar]
- An analysis of time and money spent on investigating painful total knee replacements. Article. BJMP.. 2012;5(3)
- [Google Scholar]
- Detection of aseptic loosening in total knee replacements: a systematic review and meta-analysis. Skeletal Radiol. Oct 2019;48(10):1565-1572.
- [Google Scholar]
- ACR appropriateness Criteria® imaging after total knee arthroplasty. J Am Coll Radiol. 2017/11/01/2017;14(11, Supplement):S421-S448.
- [Google Scholar]
- Recommendations for the Conduct, reporting, editing, and publication of scholarly Work in medical Journals. 2023
- [Google Scholar]
- 2023
- [Google Scholar]
- Aspiration of the knee joint before revision arthroplasty. Clin Orthop Relat Res. Oct 1996;331:132-139.
- [Google Scholar]
- Visible fluid motion on manipulation as the new threshold for intraoperatively-determined knee arthroplasty component loosening – a Delphi study. Knee Surg Sports Traumatol Arthrosc 2024
- [Google Scholar]
- The role of SPECT/CT in painful, noninfected knees after knee arthroplasty: a systematic review and meta-analysis—a diagnostic test accuracy review. J Orthop Surg Res. 2023/03/21 2023;18(1):223.
- [Google Scholar]
- MRI and SPECT/CT demonstrate, with low certainty of evidence, the highest diagnostic accuracy for aseptic knee arthroplasty loosening; a systematic comparative diagnostic test review and meta-analysis. Knee Surg Sports Traumatol Arthrosc 2024
- [Google Scholar]
- Promising results of a non-invasive measurement of knee implant loosening using a loading device, CT-scans and 3D image analysis. Clin Biomech. Apr 2023;104
- [Google Scholar]
- Weight-bearing pain and implant migration, progressive radiolucencies, radiolucency more than 2 mm, and subsidence on radiographs and CT are generally accepted criteria for knee arthroplasty loosening - an international Delphi consensus study. Knee Surg Sports Traumatol Arthrosc 2024
- [Google Scholar]
- Load-induced deformation of the tibia and its effect on implant loosening detection. Sci Rep. 2023/12/08 2023;13(1)
- [Google Scholar]
- Non-invasive quantitative assessment of induced component displacement can safely and accurately diagnose tibial component loosening in patients: a prospective diagnostic study. Knee Surg Sports Traumatol Arthrosc 2024 May 31
- [Google Scholar]