[Construction and validation of Cox proportional hazards model for the risk of hip failure after hip-preserving treatment for osteonecrosis of the femoral head].

Abstract

Objective: To explore risk factors influencing the failure of hip preservation surgery for osteonecrosis of the femoral head (ONFH), construct and validate Cox proportional hazards model for predicting the risk of postoperative failure, and present it visually as a nomogram.

Methods: A retrospective analysis was conducted on 156 patients (34 patients with bilateral lesions, 190 hips) who underwent modified core decompression surgery or multi-hole titanium cage implantation for ONFH from August 2017 to December 2023. Baseline data and follow-up information of patients were compared:(1) Demographic data, included gender, age, body mass index(BMI), and causes of ONFH. (2) Clinical scoring, modified Harris hip score (mHHS) and visual analogue scale (VAS) for pain were evaluated.(3) Imaging indicators, included ARCO staging, JIC classification, modified Kerboul angle, lateral center-edge angle (LCEA), femoral neck-shaft angle(NSA), and α angle. (4) Laboratory tests, included platelets(PT), total cholesterol(TC), triglycerides(TG), low-density lipoprotein (LDL), high-density lipoprotein(HDL), apolipoprotein-A1(Apo-A1), apolipoprotein-B(Apo-B), prothrombin time(PT), activated partial thromboplastin time (APTT), fibrinogen (FIB), and D-dimer(D-d). (5) Surgical and postoperative data, inculded surgical method, postoperative time limit for weight-bearing restriction, use of calcium supplements/anti-osteoporosis drugs after surgery, and exposure to postoperative risk factors. The imaging endpoint was defined as postoperative radiological collapse of femoral head (progressing to stage 3B or higher according to ARCO classification), and clinical endpoint was defined as undergoing total hip arthroplasty (THA). Correlations between all above variables and two types of postoperative failure outcomes were examined and analyzed by Univariate Cox regression analysis, variables with P<0.1 were included in backward stepwise regression multivariate model to screen out independent risk factors and construct a nomogram. The discrimination of model was evaluated by area under the curve(AUC) of time-dependent receiver operator characteristic(ROC) curve and C-index. The accuracy was assessed by drawing calibration curve. The clinical practicability was evaluated by decision curve analysis(DCA).

Results: All 190 hips were followed up for a period of 33.00 (21.00, 51.00) months. There were 67 hips (35.26%) with imaging failure and 32 hips (16.84%) with clinical failure. The results of univariate Cox regression analysis showed 9 variables associated with imaging failure were BMI, preoperative LDL, whether AVN CAGE was implanted, preoperative ARCO stage, JIC classification, modified Kerboul angle, PT, FIB, and postoperative restricted weight-bearing time(P<0.1);preoperative JIC classification, APTT, and FIB variables were related to clinical failure (P<0.1). The results of multivariate analysis showed preoperative ARCO stage 3A[HR=1.91, 95%CI(1.16, 3.13), P=0.011], JIC classification type C2 [HR=2.19, 95%CI(1.03, 4.65), P=0.042], modified Kerboul angle≥300° [HR=3.64, 95%CI(1.22, 10.90), P=0.021], prolonged PT [HR=1.49, 95%CI(1.14, 1.96), P=0.004], and elevated FIB(HR=1.01, P<0.001) were independent risk factors for imaging failure;JIC classification type C2(HR=8.61, P<0.001), prolonged APTT (HR=1.10, P=0.004), and elevated FIB (HR=1.01, P=0.020) were independent risk factors for clinical failure. C-index values of two models were 0.78 and 0.75 respectively. Postoperative AUC values from 1 to 3 years ranged from 0.71 to 0.84. The calibration curve fitted well, and DCA showed a relatively high clinical net benefit.

Conclusion: By constructing nomogram integrating imaging and coagulation function indicators for predicting the failure risk after ONFH hip preservation surgery, a quantitative tool could be provided for preoperative risk assessment and individualized follow-up.