This was a retrospective study based on data collected from consecutive patients undergoing primary TKA from January 2010 to December 2015. Postoperative assessments were conducted at 1 week, 3 months, 1 year, 3 years, and 5 years after surgery. This study was approved by the Medical Ethics Committee of The First Affiliated Hospital of Sun Yat-Sen University (code number  57), and all the procedures followed the principles of the Helsinki Declaration.
A total of 121 patients with a diagnosis of knee osteoarthritis (Kellgren-Lawrence classification III or IV) were enrolled in the research, 11 patients were excluded because of incomplete data. At last, 110 patient data were included in the study and analyzed. The mean age of the patients (90 females, 20 males) was 66.6±7.4. Patients were considered eligible if they met the following criteria: More than 50 years old; Reported the presence of knee pain greater than 6 months; diagnosed with knee arthritis by a surgeon; treated surgically with a KTA. Patients were excluded based on the following: varus deformity greater than 15°；rheumatoid arthritis as the primary indication for surgery; drugs that affect bone mineral density were used; Body Mass Index (BMI) was greater than 30; Failure to complete long-term follow-up.
The patients accepted unilateral primary cemented TKA by the same experienced surgeon approximately 1 week after admission in the Department of Joint Surgery of the First Affiliated Hospital of Sun Yat-Sen University. A posterior-stabilized prosthesis (DePuy Synthes, P.F.C. Sigma, Warsaw, IN, USA) was implanted for TKA.
rBMD (calibrated grayscale value) measurements
Dual X-ray absorptiometry (DXA) is the golden standard used to evaluate BMD, but it is not used in routine examinations after TKA. Hernandez-Vaquero et al reported a method based on digital X-rays images to evaluate BMD, and the consistency between the BMD measured by DXA and the relative BMD (rBMD )measured by this method was approximately 0.72 to 0.92. Therefore, this method was used to evaluate BMD in this study.
In order to ensure the comparability between the X images, all X-rays were taken on the same machine, and the patient’s knee joint was controlled in a neutral position, so as to reduce the errors caused by the difference in the position of knee. Knee flexion was minimized by fixing the tibial tubercle at the lower end of the knee. Rotation was controlled by fixing the heel and the first and second toes. Ten regions of interest (ROIs) were chosen as the measured regions in tibia: four lateral regions (L1, L2, L3, L4), four medial regions (M1, M2, M3, M4), and two distal regions (D1, D2) (Fig 1). ImageJ, (version1.8, NIH, USA) was used to measure the mean grayscale value in the established regions of the radiographs. The measured grayscale value of each designated region was calibrated using the formula:, where GC,R is the calibrated grayscale value, also representing the rBMD in a given region, GR is the grayscale value within an ROI, Ga is the value of air within the radiograph, and Gf is the grayscale value of the femoral component.
The medial tibial rBMD was defined as the mean values of M1, M2 and M3, the lateral tibial rBMD as the mean values of L1, L2 and L3, and the distal tibial plateau as the mean values of D1 and D2, and the tibial rBMD as the mean values of 10 ROIs.
Clinical data collection
Data of gender, age, BMI, Operative Duration, Kellgren-Lawrence Classification, length of Stay were obtained through patient medical records. The hip-knee-ankle (HKA) angle was measured by X-ray. All patients were clinically evaluated with respect to knee function using the Knee Society Score (KSS), the Western Ontario and McMaster University Osteoarthritis Index (WOMAC) and visual analogue scale (VAS) score. The level of activity was evaluated using the University of California Los Angeles (UCLA) Activity Rating Scale[16, 17].
All statistical analyses were performed using SPSS, Version 21.0. (SPSS Inc., Chicago, IL, USA). The Shapiro-Wilk test was used to confirm that the data were normally distributed. One-way ANOVA was used to compare clinical scores and rBMD at different time points. To verify the influence of other factors on rBMD%, Pearson’s chi-squared test was used to examine the association between sex, age, HKA angle, BMI and rBMD%. Curve fitting and nonlinear regression were performed to clarify the relationship between the UCLA activity rating and rBMD% at different time point. Post hoc power analyses were performed after the study. Significance was defined as P<0.05.