Abstract
BACKGROUND: There is a need for better diagnostic tools that identify loose total hip and knee arthroplasties. Here, we present the accuracy of different (99m)Tc-dicarboxypropandiphosphate ([(99m)Tc]Tc-DPD) SPECT/CT quantification tools for the detection of loose prostheses in patients with painful hip and knee arthroplasties. METHODS: Quantitative reconstruction of mineral phase SPECT data was performed using Siemens xSPECT-Quant and xSPECT-Bone, with and without metal artefact reduction (iMAR) of CT-data. Quantitative data (SUVmax values) were compared to intraoperative diagnosis or clinical outcome after at least 1 year as standard of comparison. Cut-off values and accuracies were calculated using receiver operator characteristics. Accuracy of uptake quantification was compared to the accuracy of visual SPECT/CT readings, blinded for the quantitative data and clinical outcome. RESULTS: In this prospective study, 30 consecutive patients with 33 symptomatic hip and knee prostheses underwent [(99m)Tc]Tc-DPD SPECT/CT. Ten arthroplasties were diagnosed loose and 23 stable. Mean-SUVmax was significantly higher around loose prostheses compared to stable prostheses, regardless of the quantification method (P = 0.0025-0.0001). Quantification with xSPECT-Bone-iMAR showed the highest accuracy (93.9% [95% CI 79.6-100%]) which was significantly higher compared to xSPECT-Quant-iMAR (81.8% [67.5-96.1%], P = 0.04) and xSPECT-Quant without iMAR (77.4% [62.4-92.4%], P = 0.02). Accuracies of clinical reading were non-significantly lower compared to quantitative measures (84.8% [70.6-99.1%] (senior) and 81.5% [67.5-96.1%] (trainee)). CONCLUSION: Quantification with [(99m)Tc]Tc-DPD xSPECT-Bone-iMAR discriminates best between loose and stable prostheses of all evaluated methods. The overall high accuracy of different quantitative measures underlines the potential of [(99m)Tc]Tc-DPD-quantification as a biomarker and demands further prospective evaluation in a larger number of prosthesis.