摘要目的:探讨激素性股骨头坏死（osteonecrosis of femoral head，ONFH）的临床与遗传风险因素。方法:建立首次使用高剂量糖皮质激素患者的前瞻性队列，纳入2015年7月至2018年1月首次接受高剂量糖皮质激素治疗的患者。患者入组前均未接受过糖皮质激素治疗，计划入组后接受等效泼尼松剂量≥30 mg/d，持续≥3周；或冲击治疗≥200 mg/d，持续≥3 d的糖皮质激素治疗。治疗前采集血液样本，行骨代谢、脂代谢指标检测；入组后第1、3、6、12和24个月行髋部MR检查，判断是否发生ONFH。研究终点为诊断ONFH，对未发生ONFH的患者随访至少2年，采用Lasso回归模型评估激素性ONFH发病的风险因素。采用1∶1匹配方法建立巢式病例对照子队列A（12例），采用全外显子组测序检测患者全血样本，筛选差异性和功能性单核苷酸多态性（single nucleotide polymorphism，SNP）/插入缺失（insertion-deletion，InDel）位点。回顾性招募诊断为激素性ONFH的患者及符合上述激素剂量要求但随访2年以上未发生ONFH的患者组成子队列B（50例），采集全血样本并使用Sanger测序技术分别对候选SNP/InDels进行外部验证。结果:入组96例，其中88例完成随访，8例（9.1%）在随访期内被确诊为激素性ONFH，开始激素治疗至确诊的时间为53.00（34.00，133.50） d。ONFH与非ONFH组患者年龄、性别和体质指数的组间差异无统计学意义；激素性ONFH组治疗首月激素剂量[32.74（29.55，47.05） mg/kg]较非ONFH组[24.00（21.10，29.45） mg/kg]更高（ Z=-2.410， P=0.016），但接受冲击治疗患者比例（分别为37.5%和10.0%）的组间差异无统计学意义（矫正χ 2=2.829， P=0.093）；激素性ONFH组载脂蛋白B/载脂蛋白A1（apolipoprotein B/ apolipoprotein A1，ApoB/ApoA1）比值0.95（0.80，1.50）较非ONFH组0.70（0.60，0.80）更高（ Z=-2.875， P<0.001）。Lasso回归模型结果提示，较高的ApoB/ApoA1比值、较低的血清Ⅰ型胶原羧基端肽β特殊序列（β-c-terminal telopeptide，β-CTX）及较高的首月激素剂量是激素性ONFH发生的三大风险因素，模型内部验证准确性为0.982。通过对子队列A进行全外显子测序发现7个差异SNP/InDels位点，在子队列B中进行验证证实携带 COLEC12突变（rs2305027，G1816A）是激素性ONFH的风险因素（ OR=6.00，95% CI：1.17，30.73）。 结论:较高的首月激素剂量、治疗前低血清β-CTX水平、高ApoB/ApoA1比例、携带 COLEC12突变（rs2305027，G1816A）可增加激素性ONFH的发生风险。

abstractsObjective:To perform a prospective cohort study to identify individual susceptibility of glucocorticoid (GC) -associated osteonecrosis of the femoral head (GA-ONFH) and their clinical and genetic risk factors. Methods:The present prospective cohort study enrolled patients who received their first GC therapy between July 2015 and January 2018 at Zhongshan Hospital. All patients did not receive any GC treatment before enrollment. Further, they planned to start GC treatment with the dose (equivalent prednisone) of ≥30 mg/d, lasted ≥3 weeks, or pulse dose ≥200 mg/d, lasted ≥3 d. Blood samples were collected before GC treatment to evaluate bone metabolism and its released factors. Hip MRI was performed at the 1st, 3rd, 6th, 12th and 24th month to diagnose GA-ONFH. All patients were followed-up for ≥2 years. The endpoint was regarded as diagnosis of GA-ONFH or completion of 2 years follow-up. Lasso regression was performed to determine which clinical features were associated with GA-ONFH. A nested case-control sub-cohort (A, n=12) was established prospectively based on the main cohort by 1∶1 matching. Whole exome sequencing was performed to screen differential and functional candidate single nucleotide polymorphisms and insertion-deletions (SNP/InDels). Another sub-cohort (B, n=50) was constructed retrospectively in patients with GA-ONFH and non-ONFH patients received standard high dose GC treatment for more than two years. The candidate SNP/InDels were verified by Sanger sequencing based on the patients from sub-cohort B. Results:A total of 96 patients were enrolled of which 88 of them (32 males and 56 females, mean age 42.30 years) completed follow-up. Eight cases (9.1%) were diagnosed with GA-ONFH. The median time from the start of GC therapy to the diagnosis of ONFH was 53.00(34.00,13.50) days. The baseline characteristics, such as age, sex and body mass index, indicated no significant difference between the ONFH group and the non-ONFH group. The cumulative GC dose of the ONFH patients in the first month was higher than that of non-ONFH [32.74(29.55, 47.05) mg/kg vs. 24.00(21.10, 29.45) mg/kg, Z=-2.410, P=0.016]. However, there was no significant difference of patients who underwent pulse therapy (37.5% vs. 10.0%, adjusted χ 2=2.829, P=0.093). The ratio of serum apolipoprotein B/apolipoprotein A1 (ApoB/ApoA1) in patients with ONFH was higher than that in non-ONFH group before GC use [0.95(0.80, 1.50) vs. 0.70(0.60, 0.80), Z=-2.875, P=0.000]. Due to the multicollinearity, Lasso regression model was performed to reduce overfitting. All variables were included in the model. The results suggested that higher ApoB/ApoA1 ratio, lower serum β-c-terminal telopeptide (β-CTX) and higher cumulative GC dose in the first month were the top three risk factors of GA-ONFH. This model had an accuracy of 0.982 in internal validation. Seven differential candidate SNP/InDels were found by whole exome sequencing of sub-cohort A. We further verified these SNP/InDels in sub-cohort B. The patients with COLEC12 mutation (rs2305027, G1816A) were at risk of GA-ONFH ( OR=6.00, 95% CI: 1.17, 30.73). Conclusion:Higher first-month GC dose, lower serum β-CTX level before treatment, higher ApoB/ApoA1 ratio and COLEC12 mutation (rs2305027, G1816A) could increase the risk of GA-ONFH.