The study was a prospective cohort study performed at a single center with 145 patients with FNFs from January 2010 to March 2012 under the approval of the Institutional Review Board. All patients were fully informed of the concept of the study, generally accepted treatment methods and the prognosis of both multiple pinning and arthroplasty before participation in the study. Among the initially enrolled 145 patients, we excluded 90 patients from the study who chose to undergo total hip arthroplasty (THA) or bipolar hemiarthroplasty (BPHA) instead. Of the 52 patients, we also excluded 22 patients due to a lack of preoperative SPECT/CT evaluations and a short follow-up length of less than two years. Finally, 30 patients were enrolled in this study, including 11 men and 19 women with a mean age of 64.3 years (range, 13 to 90 years). The average duration of follow-up was 4.0 years (range, 2.0 to 8.9 years). We used Garden's classification to classify the type of FNFs. There were 21 (56.7%) Garden I patients; four (13.3%) Garden II patients; six (20%) Garden III patients; and three (10%) Garden IV patients (21).
2. Surgical technique and postoperative management
We performed all operations within 48 hours following the initial trauma and surgically treated the patients with closed-reduction and internal fixation using three cannulated screws on a fracture table. The patients maintained partial weight-bearing using crutches or walker assistance for six weeks postoperatively. Full weight-bearing was allowed gradually after six weeks. All patients were followed-up in the clinic with serial radiographs at regular intervals (postoperative six weeks; three, six, twelve months, and annually thereafter). Since the screws used to treat fractures are made of stainless steel, we had a limitation in the detection of ONFH using magnetic resonance imaging (MRI). We, therefore, traced the occurrence of ONFH using simple radiographs (total hip anteroposterior & translateral view) of the patients. All radiographs were analyzed by a board-certified radiology specialist who was blinded to the SPECT/CT results to detect the occurrence of ONFH. The radiographical diagnostic criteria for ONFH are 1) minor or diffuse osteopenia, 2) sclerosis and/or subchondral cyst formation, 3) crescent sign and/or collapse of femoral head 4) secondary degenerative change, including osteophytes and joint space narrowing. We also confirmed the diagnosis of ONFH histopathologically by undergoing bone biopsy of the surgical specimens of patients who underwent total hip arthroplasty (THA) due to ONFH complications.
3. SPECT/CT evaluation and analysis
Preoperative SPECT/CTs were obtained within 48 hours of admission to the hospital using a dual-headed gamma camera (Infina, Hawkeye 4, GE Healthcare, Milwaukee, WI, USA) equipped with low-energy high-resolution (LEHR) collimators. The SPECT data were acquired with 60 stops over 180 at 12 s per stop using a 128 x 128 matrix within three to five hours post-injection of 1,100 MBq Tc-99m HDP. The CT acquisition was performed at 140 kvp and 2.5 mA using a 512 × 512 matrix and a 5-mm slice thickness (pitch 10, interval 2.95 mm). The SPECT images were reconstructed using an iterative algorithm provided by the manufacturer to reduce starring artifacts by bladder activity. And attenuation correction was done with CT-derived attenuation maps.
3.1 Qualitative analysis
For the qualitative analysis, two experienced board-certified nuclear medicine specialists, who were blinded from the patients' clinical information, classified the perfusion status of the femoral head on the SPECT/CT scans into two groups: the normal perfusion group and the avascular group. Inter-observer discrepancies were resolved through discussion (Fig. 1).
3.2 Quantitative analysis: head-to-head uptake ratio
We used the workstation for quantitative analysis. Scintigraphy images are subject to the amount of injected radionuclide and the variability in its distribution depending on the time interval between the injection and the time point of the test and renal and hepatic function, and differences can be seen even between individuals with normal perfusion [22-24]. Therefore, we tried to compensate for the variability of the measured value by choosing a relative method for the quantitative analysis. We first defined a 3-cm-sized three-dimensional (3-D) circle as the region of interest (ROI), of which the center was aimed at the center of the femoral head, containing maximum cancellous bone without including the cortical bone. Next, we determined the mean photon uptake of the affected femoral head by averaging the photon uptake numbers in the ROI on the coronal, axial, and sagittal planes. The mean photon uptake of the unaffected femoral head was also measured by the same method. Then, the head-to-head uptake ratio was calculated by dividing the value of the affected hip by that of the unaffected hip (Fig. 2).
3.3 Quantitative analysis: head-to-acetabulum uptake ratio
When the presence of metal implants or ONFH at a contralateral hip was confirmed, we substituted the analysis from the head-to-head ratio to the head-to-acetabulum ratio by evaluating the photon uptake numbers from the contralateral hip to the ipsilateral acetabular dome. The ROI at the acetabular dome was defined as a 1-cm-sized 3-D circle containing cancellous bone as much as possible without involving the cortex. The mean photon uptake of the acetabular dome was determined by averaging the photon uptake numbers on the coronal, axial, and sagittal planes (Fig. 3).
4. Statistical analysis
Based on the final follow-up, we divided the patients into ONFH and non-ONFH groups. The sensitivity and specificity of the qualitative evaluation of the occurrence of osteonecrosis were assessed by crossover analysis. The Mann-Whitney test was used for analyzing numerical data and Fisher's exact test was used for categorical data. A receiver operating characteristics (ROC) curve was plotted from the quantitative analysis data of two groups in order to derive a cut off value for predicting ONFH. Area under curve (AUC) of Two ROC curves were compared using Delong’s test for statistical significant change. Statistical analyses were proceeded using SPSS statistical software version 21 (IBM Co., Armonk, NY, USA), and a P-value of < 0.05 was considered statistically significant.