Data collection
This study is a retrospective analysis of all SPECT/CT arthrogram studies referred to our center between February 1, 2013, and July 1, 2018. During this period, a total of 94 SPECT/CT arthrogram studies were performed for the evaluation of persistent pain after hip and knee arthroplasty. Institutional ethics review board approval of our study protocol was obtained.
Demographic information (age, gender), type of prosthesis (cemented vs non-cemented), radiographic diagnosis, clinical findings, and surgical reports were exported for analysis. All patients had an initial standard radiograph exam followed by SPECT/CT arthrogram as part of the diagnostic pathway. Patients were only included in the study if they were subsequently followed up by an orthopedic surgeon, had an operative revision, or a minimum of 1-year period of observation. As well, joint aspiration results during the follow-up period and after revision surgery were reviewed to exclude septic etiology. Six studies were excluded from analysis due to failed tracer injections as indicated by tracer activity seen outside the joint capsule on the SPECT/CT images (i.e., extracapsular injection despite fluoroscopic guidance). These patients were typically re-booked for a repeat study and the data from these repeats may be included in the data set. Twenty-five patients were excluded since no reference standard was available, either because they did not have subsequent operative assessment or no follow-up clinic data could be obtained. The remaining 63 patient studies were included for analysis (Fig. 1).
Imaging procedure
A 22-gage needle was inserted into the joint space of interest under fluoroscopic guidance by an experienced interventional radiologist. This was approached via the inferolateral margin of femoral neck component for those with a prior hip arthroplasty or subpatellar joint space for those with a prior knee arthroplasty. Positioning within the joint space was confirmed with the injection of 2 mL of iodinated contrast (Omnipaque 300, GE Healthcare, Buckinghamshire, UK). Once confirmed, 1 mCi (37 MBq) of Tc-99m sulfur colloid in 2 mL of sterile saline was injected into the joint. The patient was then encouraged to ambulate for a minimum of 30 min before being imaged. If unable to ambulate, a minimum of 1-h delay was required before imaging.
Initially, planar images were obtained in anterior, posterior, and lateral positions (128 × 128 matrix, low-energy high-resolution collimator, 10 min per acquisition). SPECT/CT images were then acquired using a hybrid SPECT/CT system equipped with a low-energy, high-resolution collimator (Philips Precedence, Best, the Netherlands; Philips Brightview, Best, the Netherlands; or Siemens Symbia, Munich, Germany). The CT parameters were 80 mA, 120–130 kV, 512 × 512 matrix size, and 1–5-mm slice thickness. SPECT/CT was performed with a matrix size of 128 × 128, 1.0 zoom, 20 s per frame, and 120 frames at 3° intervals.
All images were reconstructed using vendor-recommended iterative reconstruction algorithms. Decay correction and attenuation correction were both applied. No post-reconstruction filter was applied.
All SPECT/CT arthrogram studies were reported by specialists licensed in nuclear medicine who were not blinded to previous imaging and clinical history. The SPECT/CT images were reviewed using Oasis workstations (Segami Corporation, Columbia, MD). SPECT/CT studies were considered negative if tracer activity was only visualized within the joint space (Fig. 2) and positive for loosening if any tracer activity was seen along the bone-prosthetic interface outside of the joint space (Figs. 3 and 4). The study was considered a failed exam if an extracapsular injection was identified on the SPECT/CT images (these patients were typically re-booked for a repeat study).
Reference standard
Patients with persistent pain following primary arthroplasty were considered to have a loosened prosthesis if it was verified at the time of operative revision or if the patient had continued pain at 1-year follow-up with progressive radiographic features of loosening. The patients were determined not to have aseptic loosening if there was no evidence of loosening at time of operative assessment, if the pain resolved during 1-year period without revision, or if the patients were subsequently diagnosed with an alternative condition that explained their symptoms. The orthopedic surgeons were aware of the patient’s clinical history and not blinded to the SPECT/CT results during the follow-up period.
Statistical analysis
Age of participants and time from primary arthroplasty were reported as mean ± standard deviation. Sex, anatomical location, and type of prosthesis were reported as ratios. The agreement between the SPECT/CT report and final diagnosis was calculated to determine the sensitivity, specificity, positive predictive values, negative predictive values, and diagnostic accuracy.