人体死后不同时间玻璃体液化学成分的变化趋势

目的测定人死后玻璃体液化学成分含量，对多种化学指标进行分析，建立各自的回归方程，使玻璃体液检测能更广泛地应用于法医实践。方法用日本产OL YMPUS AU400自动生化仪对死后时间在0．5-216h的126例尸体玻璃体液中葡萄糖等11种成分的含量进行测定。结果①玻璃体液中葡萄糖、钠、氯随死后时间的延长而逐渐下降；尿素、肌酐、尿酸、钾、钙、镁、无机磷、微量蛋白随死后时间的延长而逐渐升高。葡萄糖、钾离子和无机磷与死亡时间的相关性好（r=-0．824，0．967，0．880），而尿酸和微量蛋白则不能用于推断死亡时间（r=0．350，0．153）。②死后72h内，死亡时间的逐步回归方程为：Y=-35．15＋6．05X，R^2=0．957，X代表玻璃体液钾离子浓度。Y=-27．83＋5．49X1-1．35X2，R^2=0．960，X1、X2分别代表玻璃体液钾离子、葡萄糖浓度。Y=-16．37＋3．93X1-2．29X2＋5．36X3，R^2=0．966，X1、X2、X3分别代表玻璃体液钾离子、葡萄糖、无机磷浓度。结论①玻璃体液中葡萄糖等11种化学成分的含量随死后时间的延长而改变，其中钾离子在死后72h内与死亡时间最具有线性关系。②用逐步回归法建立多元回归方程推断死亡时间，可以提高对死亡时间估计的准确性。

Contents of Vitreous Humor of Dead Body with Different Posmortem Intervals

OBJECTIVE: To establish regression correlations between postmortem interval (PMI) and contents of human vitreous humor of dead bodies for forensic purposes. METHODS: The human vitreous humor were taken from 126 dead bodies between 0.5 to 216 hours after death, and 11 chemical elements were detected by the OLYMPUS AU400 auto-biochemistry instrument. RESULTS: (1) The glucose, natrium and chlorine in human vitreous humor decreased, while the urea, creatinine, uric acid, potassium, calcium, magnesium, phosphorus, and micro-protein increased after death. The change of glucose, potassium and phosphorus were well correlated with the PMI (r = 0.824, 0.967, 0.880). But the uric acid and micro-protein did not have a good correlation with the PMI(r = 0.350, 0.153). (2) The stepwise regression analysis established the following equations for the PMI (Y): Y = -35. 15+6.05X, R2 = 0.957 (X = potassium); Y = -27.83+ 5.49X(1) - 1.35X(2), R2 = 0.960 (X(1) = potassium, X(2) = glucose); Y = -6.37+3.93X(1) -2.29X(2) + 5.36X(3), R2 = 0.966 (X(1) = potassium, X(2) = glucose, X(3) = phosphorus). CONCLUSIONS: (1) Eleven chemical components in human vitreous humor change after death, among which postassium has the best linear correlation with the PMI within 72 hours after death. (2) The accuracy of the estimation of PMI could be improved by establishing a multi-variable equation through stepwise regression.