Nomogram to predict collapse-free survival after core decompression of nontraumatic osteonecrosis of the femoral head

Abstract

Background Nontraumatic osteonecrosis of the femoral head (NONFH) is a devastating disease, and the risk factors associated with progression into collapse after core decompression (CD) remain poorly defined. Therefore, we aim to define risk factors associated with collapse-free survival (CFS) after CD of precollapse NONFH and to propose a nomogram for individual risk prediction. Methods According to the baseline characteristics, clinical information, radiographic evaluations, and laboratory examination, a nomogram was developed using a single institutional cohort of patients who received multiple drilling for precollapse NONFH between January 2007 and December 2019 to predict CFS after CD of precollapse NONFH. Furthermore, we used C statistics, calibration plot, and Kaplan-Meier curve to test the discriminative ability and calibration of the nomogram to predict CFS. Results One hundred and seventy-three patients who underwent CD for precollapse NONFH were retrospectively screened and included in the present study. Using a multiple Cox regression to identify relevant risk factors, the following risk factors were incorporated in the prediction of CFS: acute onset of symptom (HR, 2.78; 95% CI, 1.03–7.48; P = 0.043), necrotic location of Japanese Investigation Committee (JIC) C1 and C2 (HR, 3.67; 95% CI, 1.20–11.27; P = 0.023), necrotic angle in the range of 250–299°(HR, 5.08; 95% CI, 1.73–14.93; P = 0.003) and > 299° (HR, 9.96; 95% CI, 3.23–30.70; P < 0.001), and bone marrow edema (BME) before CD (HR, 2.03; 95% CI, 1.02-4.02; P = 0.042). The C statistics was 0.82 for CFS which revealed good discriminative ability and calibration of the nomogram. Conclusions Independent predictors of progression into collapse after CD for precollapse NONFH were identified to develop a nomogram predicting CFS. In addition, the nomogram could divide precollapse NONFH patients into prognosis groups and performed well in internal validation.