Museyko O, Bousson V, Adams J, Laredo J -D, Engelke K
Institute of Medical Physics (IMP), University of Erlangen, Henkestr. 91, 91052, Erlangen, Germany.
Service de Radiologie OstéoArticulaire, Hôpital Lariboisière, Paris, France.
Osteoporos Int. 2016 Mar;27(3):1137-1147. doi: 10.1007/s00198-015-3324-6. Epub 2015 Sep 28.
For quantitative computed tomography (QCT), most relevant variables to discriminate hip fractures were determined. A multivariate analysis showed that trabecular bone mineral density (BMD) of the trochanter with "cortical" thickness of the neck provided better fracture discrimination than total hip integral BMD. A slice-by-slice analysis of the neck or the inclusion of strength-based parameters did not improve fracture discrimination.
For QCT of the proximal femur, a large variety of analysis parameters describing bone mineral density, geometry, or strength has been considered. However, in each given study, generally just a small subset was used. The aim of this study was to start with a comprehensive set and then select a best subset of QCT parameters for discrimination of subjects with and without acute osteoporotic hip fractures.
The analysis was performed using the population of the European Femur Fracture (EFFECT) study (Bousson et al. J Bone Min Res: Off J Am Soc Bone Min Res 26:881-893, 2011). Fifty-six female control subjects (age 73.2 ± 9.3 years) were compared with 46 female patients (age 80.9 ± 11.1 years) with acute hip fractures. The QCT analysis software MIAF-Femur was used to virtually dissect the proximal femur and analyze more than 1000 parameters, predominantly in the femoral neck. A multivariate best-subset analysis was used to extract the parameters best discriminating hip fractures. All results were adjusted for age, height, and weight differences between the two groups.
For the discrimination of all proximal hip fractures as well as for cervical fractures alone, the measurement of neck parameters suffices (area under the curve (AUC) = 0.84). Parameters characterizing bone strength are discriminators of hip fractures; however, in multivariate models, only "cortical" cross-sectional area in the neck center remained as a significant contributor. The combination of one BMD parameter, trabecular BMD of the trochanter, and one geometry parameter, "cortical" thickness of the neck discriminated hip fracture with an AUC value of 0.83 which was significantly better than 0.77 for total femur BMD alone. A comprehensive slice-based analysis of the neck along its axis did not significantly improve hip fracture discrimination.
If QCT of the hip is performed, the analysis should include neck and trochanter. In particular, for fractures of any type, a comprehensive slice-based analysis of the neck along its axis did not significantly improve hip fracture discrimination nor did the inclusion of strength-related parameters other than "cortical" area or thickness. One BMD and one geometry parameter, in this study, the combination of trabecular BMD of the trochanter and of "cortical" thickness of the neck resulted in significant hip fracture discrimination.
对于定量计算机断层扫描（QCT），确定了用于鉴别髋部骨折的最相关变量。多变量分析表明，具有颈部“皮质”厚度的转子骨小梁骨密度（BMD）比全髋积分BMD能更好地鉴别骨折。对颈部进行逐片分析或纳入基于强度的参数并不能改善骨折鉴别效果。
对于股骨近端的QCT，已经考虑了大量描述骨密度、几何形状或强度的分析参数。然而，在每项特定研究中，通常只使用一小部分参数。本研究的目的是从一组全面的参数开始，然后选择QCT参数的最佳子集，以鉴别有无急性骨质疏松性髋部骨折的受试者。
使用欧洲股骨骨折（EFFECT）研究的人群进行分析（Bousson等人，《骨与矿物质研究杂志》：美国骨与矿物质研究学会官方杂志26:881 - 893，2011年）。将56名女性对照受试者（年龄73.2±9.3岁）与46名患有急性髋部骨折的女性患者（年龄80.9±11.1岁）进行比较。使用QCT分析软件MIAF - Femur对股骨近端进行虚拟解剖，并分析1000多个参数，主要是在股骨颈。采用多变量最佳子集分析来提取最能鉴别髋部骨折的参数。所有结果均针对两组之间的年龄、身高和体重差异进行了调整。
对于所有近端髋部骨折以及仅对于颈椎骨折的鉴别，测量颈部参数就足够了（曲线下面积（AUC）= 0.84）。表征骨强度的参数是髋部骨折的鉴别指标；然而，在多变量模型中，只有颈部中心的“皮质”横截面积仍然是一个显著的贡献因素。一个BMD参数（转子骨小梁BMD）和一个几何参数（颈部“皮质”厚度）的组合鉴别髋部骨折的AUC值为0.83，明显优于单独的全股骨BMD的0.77。沿颈部轴线进行全面的基于切片的分析并没有显著改善髋部骨折的鉴别效果。
如果进行髋部QCT检查，分析应包括颈部和转子。特别是，对于任何类型的骨折，沿颈部轴线进行全面的基于切片的分析并没有显著改善髋部骨折的鉴别效果，纳入除“皮质”面积或厚度之外的与强度相关的参数也没有改善。在本研究中，一个BMD参数和一个几何参数，即转子骨小梁BMD与颈部“皮质”厚度的组合，对髋部骨折有显著的鉴别作用。
