Detection of aseptic loosening in total knee replacements: a systematic review and meta-analysisIntroductionTotal knee replacements are an effective surgery for osteoarthritis[1]. Last year, over 55,000 operations were performed in Australia[2]. Aseptic loosening is the most common reason for revision of primary total knee replacements in Australia and remains the major factor limiting the survival of joint replacements. Aseptic loosening is described as the failure of the integration between a prosthesis and bone in the absence of infection. Loosening reported in the first few years of implantation of a prosthesis is most likely due to failure of the implant to gain fixation, however, loosening in later years likely reflects loss of fixation due to bone resorption[2].Insidious progression of tissue damage around the implant due to particulate deposition poses a major challenge, as signs and symptoms of failure may not become clinically detectable until its late stages[3]. If osteolysis is left untreated, it can lead to periprosthetic fracture at areas of bone erosion or joint instability due to destruction of the soft tissues. These pathologies can be complex to manage and hence early identification of loosening is critical.Pre-operative evaluation and diagnosis is vital for selecting appropriative patients for revision arthroplasty, as it is major surgery with inherent complications. Sequential plain film radiography are typically used to monitor aseptic loosening over time, particularly in higher risk stratification patient groups as recommended by the Australian Arthroplasty Society[3]. Plain films are less sensitive in detecting abnormalities such as early loosening or minor implant malposition and is better at detection of gross prosthetic malposition, radiolucency and fractures[4-6]. For the assessment of loosening at a single time point, nuclear medicine techniques have been advocated[7, 8]. Their principal advantage is the elimination of metal artefact that compromises computed tomography (CT) imaging and Magnetic Resonance Imaging (MRI). This review aimed to ascertain the optimal nuclear medicine imaging modality for the diagnosis of periprosthetic loosening.The aim of this study was to compare the diagnostic accuracy of available nuclear medical imaging modalities in the detection of aseptic loosening of knee prostheses. Imaging modalities considered were bone scintigraphy; SPECT/CT (Single-photon emission computed tomography), SPECT/CT arthrography, radionuclide arthrography and PET-FDG (Photon Emission Tomography Fludeoxyglucose) scanning.Methods:Search StrategyMEDLINE, EMBASE and the Cochrane Database of Systematic Reviews were searched in accordance with PRISMA (Preferred Reporting Items for Systematic Reviews and Meta- Analyses) guidelines. This approach is advocated as an effective and efficient means to retrieving studies on diagnostic accuracy[9]. The search strategy was deliberately broad, and was constructed utilising empirical evidence to optimise the retrieval of relevant articles[10]. The Boolean operators, MeSH and keyword search terms were adapted for each electronic database and are presented in Appendix 1. All types of indexed publications written in English were considered and no date restriction was applied. The last search was completed on Friday the 28th December 2018.Study selectionIncluded studies compared the results of a single imaging modality against an appropriate criterion standard of prosthetic total knee arthroplasty loosening. Sufficient information to determine either sensitivity (number of true positives (TP) and false negatives (FN)) and/or specificity (number of false positives (FP) and true negatives (TN)) had to be available.Studies were excluded if the modality was applied after a diagnostic decision had been made. Other exclusions were studies that inseparably combined the results pertaining to knee prostheses with results at other joints (eg. total hip replacements) or where multiple pathologies (such as aseptic and septic loosening) were considered and could not be separated. Where a range of sensitivities and specificities were provided due to different diagnostic imaging criteria, the values utilised were those that achieved the “nearest to top left corner” in receiver operator curve (ROC) analysis.  Data was extracted separately by each of the two authors using a standardised data extraction tool. This sought information on population, setting, modality tested, prospective or retrospective design, blind or unblind comparison, reference standard used, number studied, and raw numbers of true positives, false positives, false negative and true negatives. Differences were resolved by consensus.Study Quality and AnalysisRisk of bias was assessed using the Quality Assessment of Diagnostic Accuracy Studies-2 (QUADAS-2) tool[11]. Both reviewers assessed each included study and differences were resolved by consensus. Risk of bias was judged as “low”, “high”, or “unclear”. If all signalling questions for a domain are answered “yes” then risk of bias was judged to be “low”. If any signalling question was answered “no” this indicated the potential for bias. In line with published recommendations, we sought to avoid utilising the unclear category other than in situations of genuine uncertainty over key signals.In line with recommendations of the Cochrane collaboration, we did not calculate pooled estimates of sensitivity and specificity[12] but generated a summary receiver operator characteristic curve to compare the diagnostic accuracy of the competing modalities[13]. Results572 studies were retrieved by the search strategy. The full text of 46 studies were retrieved and read. Of these, 12 met inclusion criteria [14-26]. The reasons for exclusion of the 34 retrieved papers are documented in Figure 1. Included studies are listed in table 1.Computerised tomographic radiography was included in the search strategy to capture the hybrid nuclear medicine imaging technique of SPECT/CT. This also provided an opportunity to confirm whether any studies had considered this radiographic modality alone. Only one such study, using cone CT, was retrieved. It used each of four studies per patient as the unit of study, precluding extraction of patient-level data[27].All included studies used operative findings, a period of clinical and radiographic observation or both as a criterion standard for aseptic loosening. We considered these to be appropriate and equivalent for the purposes of this review.The majority of studies (8 of 12) were retrospective, based on chart reviews of those patients who had been referred for nuclear medicine studies and who had subsequently undergone either surgical revision or an appropriate period of observation (Table 1).Six studies of planar bone scintigraphy were included, two of SPECT arthrography, two of radionuclide arthrography and three of FDG-PET. There was only one study relating to SPECT/CT precluding comparison to any other studies. One study[15] included considerations of 2 modalities, and two reported separately on tibial and femoral components[15, 25], so that of the 12 studies, we were able to perform 14 comparisons with the criterion standard.The sensitivities and specificities from each study for each modality are displayed in Figure 2. Most studies were small. so the confidence intervals around the point estimates of sensitivity and specificity are wide. In line with current recommendations, we have consciously not calculated pooled estimates for sensitivity or specificity.The diagnostic accuracy of the assessed modalities is displayed in the summary receiver operating characteristic curve (Figure 3). The individual curves are derived by combining the results from different studies of the same test. The curve that passes closest to the upper left corner of the graph has the greatest area under the curve, and represents the most accurate test [13], in this case SPECT arthrography.DiscussionBefore accepting these findings, it is important to consider strengths and limitations of this study. The reliability and utility of a systematic review are determined by the rigour of the techniques used in the performance of the review and the quality of the available studies. We have sought to perform this review in line with expert recommendations. Database selection and search strategies were informed by the Cochrane Collaboration[9] and supported by empirical evidence. Data was extracted from available studies by both authors, and analysed according to current recommendations, most notably the use of summary operating receiver curves.An ideal trial for inclusion would involve a prospectively applied test in a diverse and representative range of patients, the results of which are appraised blindly and compared to a definitive reference standard. For the diagnosis of aseptic arthroplasty loosening, there is no single reference standard. Most of the included studies have small numbers and were retrospective in nature. On formal assessment most had a high risk of bias. Notwithstanding these concerns, our analysis represents the current available evidence in this field.We have not considered the utility or accuracy of serial investigations, or whether some investigations would have additional utility in considering other competing diagnoses (such as FDG-PET for malignancy or infection). These considerations would be best considered in a formal decision analysis, which is outside the scope of this paper.Several studies were excluded from the current analysis on the basis that the sensitivity and specificity for individual diagnoses or joints was not able to be extracted from the data. This risks losing important data. We would recommend that future studies report findings for each combination of diagnosis, joint and modality or provide raw data that enable this information to be calculated in a 2 x 2 table.Accepting the findings of this review, nuclear arthrography with SPECT/CT provides the most accurate means of identifying aseptic prosthetic joint loosening. This observation has face value in that the pathological processes are at the implant/ bone interface, and accessing this area with a bone-avid agent directly, rather than through systemic dissemination, would seem likely to be more reliable. Moreover, the detailed spatial localisation provided by hybrid imaging enables typical, diagnostic and patterns and sites of pathological uptake to be identified. This technique has other potential advantages over non-arthrographic techniques. Septic loosening is an important differential diagnosis in this clinical setting, and as part of an arthrography study, a diagnostic arthrocentesis can be performed. This disadvantages and risks of arthrography are primarily the introduction of infection, a potentially catastrophic complication, so it is axiomatic that strict aseptic technique is applied.Time from scan to revision surgery, time of scan following index procedure, and time of revision surgery after primary surgery need to comment onIn conclusion, the available evidence suggests that the most diagnostically accurate modality for the detection of aseptic loosening in TKA is  SPECT arthrography. Although more invasive than competing techniques, the clinical consequences, including revision of the arthroplasty place a premium on correct diagnosis. These data also provide a baseline against which other imaging modalities can be compared.References:Ethgen O, Bruyère O, Richy F, Dardennes C, Reginster J-Y. Health-related quality of life in total hip and total knee arthroplasty: A qualitative and systematic review of the literature. The Journal of Bone & Joint Surgery. 2004; 86(5):963-974.Australian Orthopaedic Association National Joint Replacement Registry (AOANJRR). Hip, Knee & Shoulder Arthroplasty: 2018 Annual Report. Adelaide: AOA.Broughton N, Collopy D, Solomon M. Arthroplasty society of Australia position statement on follow-up of joint replacement pateints. Australian Orthopaedic Association 2016.Mandalia V, Eyres K, Schranz P, Toms A. Evaluation of patients with a painful total knee replacement. Journal of Bone and Joint Surgery Br. 2008; 90(3):265-271.Hirschmann1 MT, Konala P, Iranpour F, Kerner A, Rasch H, Friederich NF. Clinical value of SPECT/CT for evaluation of patients with painful knees after total knee arthroplasty- a new dimension of diagnostics? BMC Musculoskeletal Disorders. 2011; 12(36).Dennis DA. Evaluation of Painful Total Knee Arthroplasty. The Journal of Arthroplasty. 2004; 19(4 (Suppl 1)):35-40.Reinartz P. FDG-PET in patients with painful hip and knee arthroplasty: technical breakthrough or more of the same. The Quarterly Journal of Nuclear Medicine and Molecular Imaging. 2009; 53(1):41-50.Love C, Marwin SE, Palestro CJ. Nuclear medicine and the infected joint replacement. Seminars in Nuclear Medicine. 2009; 39:66-78:66-78.de Vet H, Eisinga A, Riphagen I, Aertgeerts B, Pewsner D. Chapter 7: Searching for Studies.  Cochrane Handbook for Systematic Reviews of Diagnostic Test Accuracy Version 04: The Cochrane Collaboration 2008.Deville?a WLJM, Bezemera PD, Bouter LM. Publications on diagnostic test evaluation in family medicine journals: an optimal search strategy. Journal of Clinical Epidemiology. 2000; 53:65-69.Whiting P, Rutjes A, Westwood M, Mallett S, Deeks J, Reitsma J, et al. QUADAS-2: a revised tool for the quality assessment of diagnostic accuracy studies. Ann Intern Med. 2011; 155(8):529-536.Macaskill P, Gatsonis C, Deeks J, Harbord R, Takwoingi Y. Chapter 10: Analysing and Presenting Results. In: Deeks J, Bossuyt P, Gatsonis C, eds. Cochrane Handbook for Systematic Reviews of Diagnostic Test Accuracy Version 10: The Cochrane Collaboration 2010.Jones CM, Athanasiou T. Summary receiver operating characteristic curve analysis techniques in the evaluation of diagnostic tests. Ann Thorac Surg. 2005; 79:16-20.Abele JT, Swami VG, Russell G, Masson ECO, Flemming JP. The accuracy of single Photon Emission Computed Tomography/Computed Tomography Arthrography in evaluating aseptic loosening of hip and knee prostheses. The Journal of Arthroplasty. 2015; 30:1647-1651.Chew CG, Lewis P, Middleton F, Wijngaard Rvd, Deshaies A. Radionuclide arthrogram with SPECT/CT for the evaluation of mechanical loosening of hip and knee prostheses. Ann Nucl Med. 2010; 24:735-743.Claassen L, Ettinger M, Plaass C, Daniilidis K, Calliess T, Ezechieli M. Diagnostic value of bone scintigraphy for aseptic loosening after total knee arthroplasty. Technology and Health Care. 2014; 22:767–773.Kitchener MI, Coats E, Keene G, Paterson R. Assessment of radionuclide arthrography in the evaluation of loosening of knee prostheses. The Knee. 2006; 13:220 – 225.Mandegaran R, Agrawal K, Vijayanathan S, Gnanasegaran G. The value of 99mTc-MDP bone SPECT/CT in evaluation of patients with painful knee prosthesis. Nuclear Medicine Communications. 2018; 39:397–404.Manthey N, Reinhard P, Moog F, Knesewitsch P, Hahn K, Tatsch K. The use of [18F] fluorodeoxyglucose positron emission tomography to differentiate between synovitis, loosening and infection of hip and knee prostheses. Nuclear Medicine Communications. 2002; 23:645-653.Mayer-Wagner S, Mayer W, Maegerlein S, Linke R, Jansson V, Mu?ller PE. Use of 18F-FDG-PET in the diagnosis of endoprosthetic loosening of knee and hip implants. Arch Orthop Trauma Surg. 2010; 130:1231–1238Sacchetti G, Ghisellini F, Brambilla M, Consoli AD, Fornara P, Rizzo E, et al. Quantitative Scintigraphic Evaluation of Total Knee Arthroplasties: A Feasibility Study. Clinical Orthopaedics and RElated Research. 1996; 325:181- 189.Smith SL, Wastie ML, Forster I. Radionuclide bone scintigraphy in the detection of significant complications after total knee joint replacement. Clinical Radiology. 2001; 56:221-224.Sterner T, Pink R, Freudenberg L, Jentzen T, Quitmann H, Bockisch A, et al. The role of [18F] fluoride positron emission tomography in the early detection of aseptic loosening of total knee arthroplasty. International Journal of Surgery 2007; 5:99-104.Verlooy H, Victor J, Renson L, Vandecruys A, Drent P, Mortelmans L, et al. Limitations of quantitative radionuclide bone scanning in the evaluation of total knee replacement. Clin Nucl Med. 1993; 18(8):671-674.Marx A, Saxler G, Landgraeber S, Lo?er F, Holland-Letz T, Knoch Mv. Comparison of subtraction arthrography, radionuclide arthrography and conventional plain radiography to assess loosening of total knee arthroplasty. Biomed Technik. 2005; 50:143-147.Klett R, Steiner D, Laurich S, Bauer R, Kordelle J. Evaluation of aseptic loosening of knee prostheses by quantitative bone scintigraphy. Nuklearmedizin. 2008; 47(4):163-166.Jaroma A, Suomalainen J-S, Niemitukia L, Soininvaara T, Salo J, ?ger HK. Imaging of symptomatic total knee arthroplasty with cone beam computed tomography. Acta Radiologica. 2018; 59(12):1500–1507.