Diagnosis of peri-prosthetic loosening of total hip and knee arthroplasty using 18F-Fluoride PET/CT

Periprosthetic loosening, either aseptic or induced by periprosthetic joint infection remains a major long term complication and challenge in orthopedics and trauma surgery. Sensitivity of potential loosening of the material and other causes of a painful prosthesis is essential for choosing the respective treatment option and providing the needed resources. 18F-Fluoride is a radiopharmaceutical which shows a high affinity to bone and a rapid blood clearance. The objective of this study was to assess 18F-Fluoride PET/CT´s sensitivity and specificity in diagnosing periprosthetic loosening in total hip and knee arthroplasty. We included 26 patients with 24 hip and 13 knee prostheses in our retrospective study with radiological or clinical suspicion of peri-prosthetic loosening at least one year after implantation. Results of 18F-Fluoride PET/CT imaging were compared with surgical results or clinical follow-up if surgery was not performed. On the basis of our data we found a sensitivity of 95.00 %, a specificity of 87.04 % and an accuracy of 89.19 % for 18F-Fluoride PET/CT. The results of our study show that 18F-Fluoride PET/CT is a useful and promising technique in diagnosing periprosthetic loosening of total hip and knee arthroplasties. Further investigation should focus on different uptake patterns of the isotope in periprosthetic joint infection and therefore distinguishing aseptic from septic loosening and enhancing the diagnostic value of this imaging method.


INTRODUCTION
Total hip and knee arthroplasties in end-stage arthritis contribute an essential number to today´s orthopedic procedures. Periprosthetic loosening, either aseptic or induced by periprosthetic joint infection remains a major long term complication and challenge in orthopedics and trauma surgery. According to estimations between 0.4 and 4% of joint replacements will face periprosthetic infection and 2-18 % aseptic loosening [1]. When dealing with painful prostheses an early distinguishment between potential loosening of the material and other causes is essential for choosing the respective treatment option and providing the needed resources [2].
According to current understanding a chronic inflammatory process of the bone-prosthesis or cementprosthesis interface causes implant loosening. Wear particles are being phagocytosed by macrophages which induce chronic inflammation and osteolysis with activation of osteoblastic and osteoclastic activity [3,4]. Early before the gross implant motion sets in and radiographic changes can be recognized, so called micromotion already causes pain and makes diagnosis difficult. Today´s standard diagnostic procedures for ruling out prosthetic loosening most often include x-ray and bone scan imaging, as well as joint punctures and laboratory tests which feature wide ranges in specificity and sensitivity [1]. Especially plain radiographs often show signs of loosening only after a long delay of time. Traditional bone scan imaging comes along with a low spatial resolution. 18 F-Fluoride PET/CT provides a promising alternative. 18 F-Fluoride is a radiopharmaceutical which shows a high affinity to bone and a rapid blood clearance with high bone-to-background ratio in a shorter time than for standard 99m Tc-based tracers [5,6]. In combination with positron emission tomography and its excellent spatial resolution 18 F-Fluoride PET/CT may offer highly valuable images for detecting loose components of total hip and knee arthroplasties.
Therefore, the objective of this study was to assess 18 F-Fluoride PET/CT´s sensitivity and specificity in diagnosing periprosthetic loosening in total hip and knee arthroplasty.

RESULTS
The collective of 26 patients included 15 male and 11 female patients with a mean age of 67.5 years (50.1 -92.2 years). We evaluated a total of 24 hip and 13 knee prostheses. Since the high resolution of the imaging allows for a separate evaluation of the femoral, tibial and acetabular part of an arthroplasty we were able to analyze 74 prosthetic components. The mean time between implantation and 18 F-Fluoride PET/CT scan was 6.5 years (range 1.0 -19.6 years).
The clinically silent additional 11 prostheses (22 components) were counted as true negative since the 18 F-Fluoride PET/CT showed no signs of loosening.
For 18 patients and therefore 18 prostheses (36 components, 48.6%) the 18 F-Fluoride PET/CT diagnosis was confirmed or disproved by intraoperative findings. In these cases the surgeon declared the respective components as loose or stable as stated in his operational report.
For 19 prostheses (38 components, 51.4%) without indication for revision surgery a clinical follow up lasting at least 6 months confirmed the 18 F-Fluoride PET/CT diagnosis of a loosened or non loosened prosthesis.
One patient did not undergo revision surgery of two total knee arthroplasties due to age and the high risk profile of the operation although showing distinctive signs of prosthetic loosening on both sides. The clinical und radiological follow up for one year confirmed the diagnosis of loosened prostheses and its progression. These cases were therefore evaluated as true positive.
One patient with a positive 18 F-Fluoride PET/CT finding for a loosening of the tibial component of the total knee arthroplasty has not been operated and showed no clinical symptoms in the follow up counting as false positive. 18 F-Fluoride PET/CT correctly identified 18 out of 19 loosened total hip and knee arthroplasty components. In 6 out of 55 components without intraoperative or clinical signs of loosening we found a positive 18 F-Fluoride PET/ CT enhancement around the prosthesis counting as false positive. The others were correctly identified as not loosened in the scan.
On the basis of these data and considering every single prosthesis component (total 74, view Table 1) we found a sensitivity of 95.00 %, a specificity of 87.04 % and an accuracy of 89.19 %. These data include the clinically silent additional 11 prostheses (22 components).
When considering the respective endoprosthesis as one compound without distinguishing the femoral, tibial or acetabular component (n=37, view

DISCUSSION
The aim of this study was to evaluate the role of 18 F-Fluoride PET/CT in diagnosing periprosthetic loosening in total hip and knee arthroplasty.
M. Blau et al. were the first to use Fluorine-18 in bone imaging in 1962 [7], whereas Creutzig et al. presented a study of 31 hip prosthesis which he examined towards loosening and infection using planar 18 F-Fluoride scanning in 1976 [8]. Since then little research has been performed regarding this radiopharmaceutical and prosthesis loosening. Nevertheless, other impairments of bone metabolic activity have been very well investigated using 18 F-Fluoride PET, including renal osteodystrophy [9], osteoporosis [10], osseointegration of bone allografts [11], femoral head necrosis [12] and even bone malignancies [13].
In 2006 Sterner et al. showed in a study of 14 knee prostheses a sensitivity of 100% for diagnosing aseptic loosening with 18 F-Fluoride PET/CT, whereas specificity was very low with 56% and the relatively high rate of false positive results could not be explained by the authors [14]. Still, they were the first to present results of 18 F-Fluoride PET imaging in diagnosis of aseptic loosening. Compared to the above mentioned study our 18 F-Fluoride PET/CT data show both a high sensitivity and superior specificity.
The so far most investigated radionuclide procedure used for imaging joint arthroplasties is bone scintigraphy with Technetium-99m ( 99m Tc) labeled diphosphonates, usually methylene diphosphonate (MDP). No matter what www.oncotarget.com      type of protocol or evaluation has been used, the accuracy of the method never exceeded 70%, whereas specificity has always been reported low, mostly due to the limitations in spatial resolution and therefore evaluation of distinctive uptake patterns of the isotope [15][16][17][18]. 18 F-Fluoride PET/ CT however shows an excellent resolution enabling a more accurate observation of the pattern resulting in a higher specificity of the diagnosis. In 2011 Choe et al. presented a comparatively large prospective trial towards differentiation between septic and aseptic loosening by using 18 F-Fluoride PET/ CT including 49 patients with 65 total hip arthroplasties. To our knowledge this represents the first prospective investigation in this regard [19] and the highest number of participants. The method involved measurement of the degree of 18 F-Fluoride uptake in periprosthetic tissue for septic and aseptic loosening. Our study did not include these types of measurements, but since differentiation between aseptic and septic loosening is key to an adapted treatment, further investigation in this matter is warranted.
Our and the above mentioned studies included patients whose prostheses have been implanted at least 1 year prior to 18 F-Fluoride PET/CT imaging. According to Creutzig et al. unspecific 18 F-Fluoride uptake especially around the acetabular component of total hip arthroplasty fades out about 9 months after implant surgery [8]. This is an argument in favor of using 18 F-Fluoride for this indication, considering that fluorodeoxyglucose ( 18 F-FDG), the most commonly used PET tracer, shows non-specific periprosthetic increased uptake for as long as several years, even in patients without evidence of infection or loosening [20]. However, also 18 F-Fluoride may show persisting non-specific uptake at the prosthesis/ bone interface of total hip and total knee arthroplasties [Son et al.], which necessitated the definition of nonspecific uptake patterns of 18 F-Fluoride in our study to reduce the number of false-positive results.
The emphasis in our study was put on the differentiation between loosened and not loosened prostheses, the differentiation between septic and aseptic loosening was not in the focus of our study. A quantitative evaluation has not been performed so far and serves as an limitation of the current study. However dynamic quantitative evalutation might give an insight into differentiation between septic and aseptic loosening. To address this topic, the evaluation of dynamic 18 F-Fluoride PET/CT scans with emphasis on an infection-related loosening will be part of further ongoing studies.

Patient population
The study was approved of by the local institutional review board.
We included 26 patients in our retrospective study who suffered from painful total hip or total knee arthroplasties with radiological or clinical suspicion of peri-prosthetic loosening at least one year after implantation. Patients with prior oncologic disease have been excluded from the study. In the case of multiple inlaying prostheses without pain or clinical suspicion of loosening, the additional total knee and hip arthroplasties were also included into the study since the 18 F-Fluoride PET/CT also produced images of these implants. All included patients were examined via 18 F-Fluoride PET/ CT between September 2014 and July 2017.
All patients were subject to routine clinical examination, laboratory and radiological studies. Data, such as development of pain and other clinical symptoms were gathered by chart review of medical records. No definite diagnosis had been established in any patient before the 18 F-Fluoride PET/CT scan was performed. Revision operations were performed by experienced surgeons 1.4 ± 0.2 months after PET/CT scan. If no operation has been performed, the final evaluation and diagnosis was based on the long term clinical follow up of minimum 6 months (median: 11.3 months, range 6.0 -29.7 months). In these cases the patient either refused to undergo revision surgery and showed further progression of the radiological signs of loosening or experienced spontaneous regression of symptoms. All patients were examined with informed consent. Both components of the total hip (acetabular and femoral part) and total knee arthroplasty (femoral and tibial part) were independently evaluated as true positive or true negative according to the intraoperative findings or longterm clinical follow up. 9 patients had two, and one patient had 3 inlaying prostheses.

PET scanning
PET imaging was performed with a Biograph 2 PET/ CT scanner (Siemens Healthcare GmbH, Erlangen, Germany) with an axial field of view of 16.2 cm and a transversal field of view of 58.5 cm. 18 F-Fluoride was injected intravenously at a mean activity of 292 (± 24) MBq. Static emission scans of the legs were performed at a mean of 104 (± 18) min after injection (4 minutes emission time per bed position, 3D mode). PET images were reconstructed iteratively (4 iterations, 8 subsets, 256 × 256 matrix, 5mm slice thickness). A low-dose CT (16 mAs, 130 kV) was acquired for attenuation correction and anatomical correlation.

Image interpretation
The scans were visually evaluated by an experienced nuclear medicine physician. The evaluation criteria were based on the pattern and location of the 18 F-Fluoride uptake at the prosthesis/bone interface or cement/bone interface.
The criteria we used in our current study to differentiate non-specific uptake of 18 F-Fluoride from patterns specific for loosening are in close correlation with uptake patterns described by Son et al., who previously examined postoperative 18 F-Fluoride uptake in asymptomatic patients up to 25 months after hip or knee arthroplasty [21]. Our criteria for non-specific uptake correspond to areas of the bone interface which showed intense uptake in more than 75% of asymptomatic patients in the study by Son et al. (Normal or no uptake see Figure  1 (A,D)) Specifically, these are the cranial and caudomedial parts of the acetabular component and the distal tip of the femoral component of hip prostheses. Additionally, we categorized small areas of focal uptake in the proximal region of the femoral component as non-specific, as an increased frequency of persistent uptake of 99m Tc-HDP has been observed in these areas in a previous study in asymptomatic patients with uncemented hip prostheses by Kim et al. [22], see Figure 1B. Regarding total knee arthroplasty, in correlation with the data by Son et al., uptake along the dorsal and caudal parts of the femoral sleds was rated as non-specific, as well as uptake solely in the horizontal surfaces in the proximal region of the tibial component. These criteria also closely match the methods used in the study by Sterner et al. for the detection of aseptic loosening of total knee arthroplasties by 18 F-Fluoride PET [14]. Additionally, we categorized uptake in the patellar region as non-specific, as well as small focal uptake at the tip of a pin/shaft of the femoral component, see Figure 1E.
Uptake exceeding these areas was interpreted as positive for loosening. Regarding hip prostheses, circular uptake at the interface encompassing more than half of the acetabular component was interpreted as positive for loosening, whereas for the femoral component, extended uptake around the trochanter regions or the distal parts of the shaft (exceeding the area of the tip) was rated positive ( Figure  1C). With regard to total knee arthroplasties, uptake at the interface encompassing more than half of a condylar sled extending into the anterior parts, or extensive uptake around the pin/shaft of the femoral component was categorized as suspicious for loosening, as well as tracer accumulation around the pin/shaft of the tibial component ( Figure 1F).

Data analysis
The analogically recorded values out of the register were digitized and exported to MS Excel 2007 (Microsoft, Redmond, WA, USA) and SPSS 17 (IBM, Armonk, NY, USA) for further calculation. Subject of calculation by 2x2 table were sensitivity, specificity, accuracy.

CONCLUSIONS
The results of our study show that 18 F-Fluoride PET/ CT is a useful and promising technique in diagnosing periprosthetic loosening of total hip and knee arthroplasties. It demonstrates a very high sensitivity and specificity, not only compared to those of other imaging tools such as traditional bone scan, but also to other published trials in this regard. Moreover, it shows potential to completely replace traditional bone scan in the diagnostic pathway of periprosthetic loosening. Due to the promising findings of this retrospective analysis, further prospective studies are warranted with the recruitment of a higher number of patients to define the role and usage of 18 F-Fluoride PET/CT in routine clinical practice. Future studies should also focus on the uptake kinetics of 18 F-Fluoride in periprosthetic joint infection to assess its value in distinguishing aseptic from septic loosening. Furthermore, the PET-scanning technique is also applicable with other tracers, such as 68 Ga-zoledronate, which shows a slightly better availability than 18 F-Fluoride and is therefore subject of a current prospective study of the same group.