【作者】 袁东；

【导师】 余卫；

【作者基本信息】 中国协和医科大学 ， 影像医学与核医学， 2007， 博士

【摘要】 第一部分女性股骨近端几何力学参数与髋部骨折的相关性研究目的探讨女性股骨近端几何力学参数随年龄及骨密度的变化及其对骨强度的影响和预测髋部骨折危险性的价值。材料与方法共收集髋部骨折的女性73人,年龄45～84岁,平均68.7±9.2岁;对照组女性625人,年龄45～85岁,平均66.6±9.3岁。进行双能X线股骨近端骨密度测量,骨折组选取骨折对侧股骨近端,对照组选取右侧股骨近端进行测量。在分析骨密度的同时,由HSA（Hip strength analysis）软件在DXA扫描图像基础上得出股骨近端几何力学参数,包括股骨颈横截面积（Cross-sectional area,CSA）、股骨颈横截面转动惯量（Cross-sectional moment of inertia,CSMI）、股骨颈宽度（Femoral neck width,FNW）、髋轴长（Hip axis length,HAL）、股骨颈干角（Neck shaft angle,NSA）及股骨颈强度指数（Femur strength index,FSI）。结果CSA、CSMI及FSI与年龄呈负相关,相关系数分别为（r=-0.463,P＜0.01;r=-0.277,P＜0.01;r=-0.087,P＜0.05）,CSA、CSMI及FSI在骨质疏松组低于骨量正常组,骨质疏松组FNW大于骨量正常组。CSA、CSMI及FSI在骨折组低于对照组,FNW在骨折组高于对照组,HAL在骨折组高于对照组,NSA在骨折组低于对照组。经Logistic回归分析,FSI、CSMI联合T分数预测髋部骨折危险性的ROC曲线下面积高于单独T分数的ROC曲线下面积（ROC曲线下面积分别为0.794,0.682,P＜0.01）。结论在老龄化过程中,股骨近端骨强度的减低不仅与骨量损失有关,而且与其几何力学因素相关。T分数联合CSMI及FSI可以提高对髋部骨折危险性预测的效能。第二部分女性股骨近端磁共振波谱成像与骨质疏松的相关性研究目的探讨¹H MRS不同测量指标与DXA所测骨密度的相关性及其对骨质疏松的评估价值。材料与方法共收集112例女性志愿者（32～79岁,平均62.1±10.4岁）分别进行以下检查:右股骨近端双能X线（DXA）骨密度测量;右股骨近端常规扫描斜冠状位T₁WI、T₂WI及斜矢状位T₁WI。采用单体素PRESS（Point resolved spectroscopy）序列在斜冠状位平面上分别采集股骨颈、转子间及大转子¹H波谱,相应谱线采用SAGE 7.0软件分析并计算相关MRS定量指标,包括脂水比（Lipid/water ratio,LWR）,脂肪含量（Fat fraction,FF）,脂峰线宽（Line width,LW）,不饱和指数（Unsatuaration index,UI）。每例受检者上述检查项目均在一日内完成。结果股骨颈及转子间LWR及FF均与年龄呈低度正相关（LWR:股骨颈r=0.297,转子间r=0.393,P＜0.01;FF:股骨颈r=0.358,转子间r=0.378,P＜0.01）。在经骨密度校正后,相关性仍然具有统计学意义（LWR:股骨颈r=0.205,转子间r=0.233,P＜0.05;FF:股骨颈r=0.235,转子间r=0.253,P＜0.05）。本研究显示股骨颈、股骨转子间及大转子LWR及FF均与骨密度呈中到低度负相关,相关系数分别为（LWR:-0.482,-0.401,-0.419,P＜0.01;FF:-0.462,-0.333,P＜0.01;-0.207,P＜0.05）。在经年龄校正后,股骨近端LWR与骨密度相关性仍然存在,三个部位相关系数分别为（-0.397,-0.341,-0.344,P＜0.01）。股骨颈及转子间FF与骨密度相关性经年龄校正后仍存在负相关（r分别为-0.415,P＜0.01;-0.313,P=0.01）,股骨大转子FF在经年龄校正后与骨密度未见相关性（r=-0.177,P=0.063）。本研究显示绝经后组LWR及FF较绝经前组增高。股骨颈及转子间LWR和FF在骨量正常组低于骨量减少及骨质疏松组,而骨量减少组和骨质疏松组间未见统计学差别。大转子LWR只在骨量正常组与骨质疏松组、骨量减少组与骨质疏松组间存在统计学差异,而在骨量正常与骨质减少组间未见统计学差别。FF在大转子区域与骨量变化未见相关性。股骨颈及转子间脂峰LW与年龄呈低度负相关（r分别为-0.302和-0.255,p＜0.01）,但是当将BMD引入相关分析后,两者相关性则不再具有统计学意义（股骨颈LW:r=-0.033,p=0.730,转子间LW:r=-0.031,p=0.748）。脂峰LW在绝经前组要高于绝经后组。股骨近端三个测量部位的脂峰LW均与BMD呈中度正相关（相关系数分别为股骨颈r=0.550,P＜0.01;转子间r=0.417,P＜0.01;大转子r=0.409,P＜0.01）,当控制年龄因素后,相关性仍有统计学意义（相关系数分别为股骨颈r=0.448,P＜0.01:转子间r=0.350,P＜0.01;大转子r=0.396,P＜0.01）。本研究显示股骨颈及转子间UI随年龄增大呈现下降趋势（r分别为-0.374和-0.334,p＜0.01）,经BMD校正后相关性仍然具有统计学意义（股骨颈r=-0.258,P=0.005;转子间r=-0.215,P=0.025）。股骨颈及转子间UI与骨密度呈正相关（r分别为0.281,0.308,p＜0.05）,经年龄校正后,相关性依然存在（股骨颈r=0.203,P=0.021;转子间r=0.211;P=0.014）。大转子UI与年龄和BMD未见相关性。股骨颈及转子间UI在绝经前组高于绝经后组。股骨颈和转子间UI在骨量正常组要高于骨质疏松组,在骨量正常组与骨量减低组及骨量减低组与骨质疏松组间UI均未见统计学差异。大转子区域UI三个组间未见差别。右股骨近端三个测量部位LWR、FF及脂峰LW诊断骨质疏松ROC曲线下面积大小为0.728～0.846,股骨颈LWR、FF及脂峰LW诊断骨质疏松ROC曲线下面积高于转子间和大转子。结论股骨近端骨髓脂肪含量随年龄增长而逐步增加,股骨近端脂肪含量随骨密度减低而有逐渐增高的趋势。脂峰LW与年龄相关性不大,而与骨密度呈正相关。股骨近端骨髓中不饱和脂肪酸所占比例随年龄增长逐渐下降,并随骨密度减低而降低。LWR、FF及LW可以作为无创性评估骨密度的磁共振指标。第三部分女性股骨近端R2*值与骨质疏松的相关性研究目的探讨右股骨近端R2*测值与DXA所测骨密度的相关性及其对骨质疏松的评估价值。材料与方法共收集112例女性志愿者（32～79岁,平均62.1±10.4岁）分别进行以下检查:右股骨近端双能X线（DXA）骨密度测量;右股骨近端常规扫描斜冠状位T₁WI、T₂WI及斜矢状位T₁WI,并扫描斜冠状位MEGRE（Multi-echo gradient echo sequence）序列。利用R2*测量软件在最大斜冠状位平面上分别测量股骨颈、转子间及大转子R2*值。结果股骨颈、转子间及大转子R2*值与年龄呈中到低度负相关（r分别为-0.521,-0.285,-0.399,p＜0.01）,当引入骨密度作为协变量进行相关分析,股骨颈及转子间R2*与年龄相关性不再具有统计学意义,大转子R2*值与年龄仍呈低度负相关,相关系数减小（r=-0.199,p=0.030）。股骨近端三个测量部位R2*值与DXA所测骨密度呈中到高度正相关（股骨颈r=0.701,P＜0.01;转子间r=0.492,P＜0.01;大转子r=0.550,P＜0.01）,经年龄因素校正后,相关性依然有统计学意义（股骨颈r=0.572,p＜0.01,转子间r=0.424,p＜0.01,大转子r=0.453,p＜0.01）。在绝经前组股骨近端三个测量部位的R2*值均高于绝经后组。在不同骨量组间的协方差分析中,股骨近端三个测量部位间的两两比较均具有统计学意义。股骨颈R2*值判定骨质疏松ROC曲线下面积为0.859,转子间为0.767,大转子为0.851。结论右股骨近端R2*值随骨密度下降有逐渐减低的趋势,与年龄相关性意义不大。右股骨近端R2*值可以作为无创性判定骨质疏松的指标,股骨颈R2*值判定骨质疏松的ROC曲线下面积最大。更多还原

【Abstract】 PART ONE:Correlation of structural geometric properties of the proximal femur with hip fracture in womenObjective:To evaluate the influence of age and bone mineral density on geometric structure of the proximal femur in women;to study the influence of hip geometry on hip strength and its value in predicting hip fracture.Material and methods:DXA measurements of the proximal femur were obtained from 698 women,73(age: 68.7±9.2) with prior hip fracture and 625(age:66.6±9.3) controls.In addition to the conventional densitometry measurements,structural variables were determined using the Hip Strength Analysis(HSA) program,including cross-sectional area(CSA), cross-sectional moment of inertia(CSMI),femoral neck width(FNW),hip axis length(HAL),neck shaft angle(NSA) and femur strength index(FSI).Results:CSA, CSMI and FSI were negatively correlated with age(r=-0.463,P＜0.01;r=-0.227, P＜0.01;r=-0.087,P＜0.05;respectively).CSA,CSMI,FSI were significantly lower and FNW significantly higher in osteoporotic group compared with healthy group. CSA,CSMI,FSI were significantly lower and FNW significantly higher in the fracture group compared with controls.HAL was significantly longer and NSA significantly narrower in hip fracture group compared with controls.The fracture discrimination ROC curve for the logistic regression probability model incorporating FSI and CSMI was significantly larger than that for T score alone(0.794 vs.0.682, P＜0.01),indicating that fracture discrimination improved when these femur structural variables were combined with BMD measurements.Conclusion:The reduction of the proximal femur strength in aging is not only influenced by bone loss,but also by the changed properties of its hip geometric structure.T score combining with CSMI and FSI may have a more effective capability of predicting hip fracture.PART TWO:Pilot study of correlation between MRS and osteoporosis in proximal femur in womenObjective:The purpose of this research is to investigate the correlation between MRS variables and BMD in proximal femur;to evaluate the diagnostic value of MRS variables for osteoporosis.Materials and methods:Proton MRS and BMD measurement of femoral neck,intertrochanteric and trochanter region of the right proximal femur were performed in 112 female subjects(age:62.14±10.4)。The single voxel PRESS(Point resolved spectroscopy) sequence was used at 3.0T MR imaging. All data were analyzed at SAGE 7.0 program and LWR(Lipid/water ratio),FF(Fat fraction),lipid line width(LW) and lipid unsaturation index(UI) were calculated. BMD measurements of the hip were obtained with DXA.Results:LWR and FF of the neck and intertrochanteric region had significant correlation with age(neck:LWR: r=0.293,FF:r=0.397,P＜0.01;intertrochanteric:LWR:r=0.358,FF:r=0.378,P＜0.01). The correlation of LWR and FF of the two measured regions in proximal femur with age remained significant after adjustment for BMD(neck:LWR:r=0.205,FF:r=0.233, P＜0.05;intertrochanteric:LWR:r=0.235,FF:r=0.253,P＜0.05).LWR and FF of the three measured regions were negatively correlated with BMD(LWR:r=-0.482,-0.401, -0.419,P＜0.01;FF:r=-0.462,-0.333,P＜0.01;r=-0.207,P＜0.05),and the significance remained after adjustment for age except for FF of trochanter region(LWR: r=-0.397,-0.341,-0.344,P＜0.01;FF:r=-0.415,P＜0.01;r=-0.313,P--0.01;r=-0.177, P=0.063).LWR and FF had significant difference between pre-and postmenopausal subgroups with these variables higher in premenopausal women.LWR and FF were significantly lower in healthy group compared with osteopenic and osteoporotic group for neck and intertrochanteric region.No significant difference was observed between osteopenic and osteoporotic groups in these two regions.LWR for trochanter was significantly higher in osteoporotic group compared with healthy and osteopenic groups while no significant disparity was obtained between healthy and osteopenic groups.FF for trochanter had no significant disparity in all three subgroups.Lipid LW of neck and intertrochanteric region was negatively correlated with age(r=-0.302, -0.255,p＜0.01),but the correlation became insignificant after adjustment for BMD(neck region:r=-0.033,p=0.730,intertrochanteric region:r=-0.031,p=0.748). Lipid LW in premenopausal group was significantly higher compared with postmenopausal group.Lipid LW of three measured regions were positively significant with BMD(neck region:r=0.550,P＜0.01;intertrochanteric region:r=0.417, P＜0.01;trochanter region:r=0.409,P＜0.01),and the correlation still remained significant after adjustment for age(neck region:r=0.448,P＜0.01;intertrochanteric region:r=0.350,P＜0.01;trochanter region:r=0.396,P＜0.01).UI of neck and intertrochanteric region were negatively significant with age factor(r=-0.374,-0.334, p＜0.01;respectively) and the correlation remained significant after adjusted for BMD factor(neck region:r=-0.258,P=0.005;intertrochanteric region:r=-0.215,P=0.025).It showed positive correlation between UI of neck and intertrochanteric region and BMD factors(r=0.281,0.308,p＜0.05;respectively) and the correlation remained significant after the age factor was controlled as one partial variable(neck region: r=0.203,P=0.021;intertrochanteric region;r=0.211;P=0.014).It showed no significant correlation between UI of trochater region and age as well as BMD.UI of neck and intertrocanteric region were significantly lower in postmenopausal group compared with premenopausal group.UI of neck and intertrochanteric region were significantly lower in osteoporotic group compared with healthy group.There were no significant disparity between healthy and osteopenic group as well as osteopenic and osteoporotic group for UI of neck and intertrochanteric region.No significant difference was observed for UI of trochanteric region among all three groups.ROC curve area of LWR,FF,lipid LW of the three measured regions for osteoporosis discrimination ranged from 0.728～0.859.ROC curve area of LWR,FF and lipid LW for neck region were larger compared with those of introchanteric and trochanter region.Conclusion:The marrow fat content of proximal femur is positively correlated with age and negatively correlated with BMD.Lipid LW was positively correlated with BMD and has no significant correlation with age.UI of proximal femur has negative correlation with age and positive correlation with BMD.LWR,FF and lipid LW can be used as noninvasive MR variables in evaluation of BMD as well as in determination of osteoporosis.PART THREE:Assessment for osteoporosis with R2~* characteristics in proximal femur in womenObjective:To investigate the correlation of R2* measurement with BMD obtained from DXA in proximal femur,to evaluate the diagnostic value of R2~* for osteoporosis. Material and methods:T2~* map of the proximal femur were obtained at 3.0T MR in 112 women(age:62.1±10.4).A multi-echo gradient echo sequence called MEGRE was used in this study.R2~* values were measured in three regions of interest:femoral neck,intertrochanteric region and trochanter.BMD measurements of the hip were obtained with DXA.Results:The R2~* values of all three regions were negatively correlated to age(femoral neck region:r=-0.521,P＜0.01;intertrochantetic region: r=-0.285,P＜0.01;trochanter region:r=-0.399,P＜0.01).After adjusted for BMD, insignificant correlation was obtained on femoral neck region and intertrochanteric region,while trochanter region still showed significant correlation to age,only to have a lower correlation coefficient(r=-0.199,P＜0.030).The R2~* value of all three measurement regions were significantly correlated to BMD(neck region:r=0.701, P＜0.01;intertrochantetic region:r=0.492,P＜0.01;trochanter region:r=0.550,P＜0.01), even so after adjustment for age(neck region:r=0.572;p＜0.01,intertrochantetic region:r=0.424,p＜0.01;trochanter:r=0.453,p＜0.01).Mean R2~* values were significantly higher in premenopausal group compared with postmenopausal group. Significant difference of R2~* values between all three groups as regard to healthy women,osteopenic women and osteoporotic women were obtained in three proximal femur measurement regions.The ROC curve area for osteoporosis determination was 0.859 for neck region,0.767 for intertrochanteric region and 0.851 for trochanteric.Conclusion:The R2~* values of proximal femur were significantly correlated with BMD,but not to age,indicating that it is a MR variable for noninvasive detection of changes in BMD.The R2~* value of neck region has the largest ROC curve area for discrimination value for osteoporosis compared with intertrochanteric region and trochanter.更多还原