血糖水平对糖尿病患者肾小球滤过率估算公式的影响

目的 评价糖尿病患者血糖水平对肾小球滤过率（GFR）公式估算结果的影响；比较不同血糖水平Cockcroft-Gault（CG）公式和MDRD公式法估算GFR对诊断肾功能不全的差异。 方法 选取1210例糖尿病患者，同步检测99mTc-GFR（iGFR）、Scr和糖化血红蛋白（HbA1c）。运用CG和MDRD公式计算GFR估计值（eGFRCG、 eGFRMDRD）。依据肾脏病透析预后质量指南（K/DOQI）的建议将糖尿病患者分为iGFR正常组589例NGFR组，iGFR≥90 ml&#8226;min-1&#8226;(1.73 m2)-1]，iGFR轻度下降组GGFR组，60≤iGFR<90 ml&#8226;min-1&#8226;(1.73 m2)-1]470例，iGFR中度下降组MGFR组，30≤iGFR<60 ml&#8226;min-1&#8226;(1.73 m2)-1]151例。根据HbA1c的四分位点（7.1%，10.5%）分为4组（<7.1%、7.1%~8.6%、8.7%~10.4%、≥10.5%），其中HbA1c<7.1%者定义为血糖控制较好组，HbA1c≥10.5%定义为血糖控制差组。采用Spearman相关分析、t检验、Bland-Altman分析、受试者工作特征（ROC）曲线等评估方程的偏离度、准确度，以及血糖对估算结果的影响。 结果 eGFRMDRD在各GFR亚组中均高估GFR；eGFRCG在NGFR组中低估GFR，差异有统计学意义。Bland-Altman分析结果显示，血糖控制较差组的eGFRMDRD的偏差高于血糖控制较好组的eGFRMDRD；血糖控制较差组的eGFRMDRD15%和30%准确性低于血糖控制较好组的eGFRMDRD，差异有统计学意义。血糖控制较差组和较好组间eGFRCG偏差及准确性差异均无统计学意义；而eGFRCG的偏差高于eGFRMDRD，差异有统计学意义。血糖控制良好组CG公式和MDRD公式在诊断肾功能不全患者的ROC曲线下面积差异无统计学意义。血糖控制较差组eGFRMDRD ROC曲线下面积显著大于eGFRCG曲线下面积，差异有统计学意义。 结论 糖尿病患者采用MDRD和CG公式法可导致GFR估计差误。MDRD公式的eGFR估计值受到血糖的影响较大，MDRD公式法高估GFR。MDRD公式在血糖控制较差的患者对肾功能不全患者的估算效应要优于CG公式。

Impact of glucose level on estimated glomerular filtration rate in diabetic patients

Objective To investigate the influence of blood glucose level on the estimated glomeruhr filtration rate(eGFR)by Cockcroft-Gault(CG)and Modification of Diet in Renal Disease(MDRD)formula in diabetic patients,and to explore the difference between CG and MDRD formula to estimate GFR in different levels of blood glucose in the diagnosis of moderate renal insufficiency.Methods A total of 1210 diabetic patients(650 males and 560 females)were enrolled in this study.HbAlc,Scr and isotopic GFR(iGFR)((99m)~ Tc-DTPA)were measured.CG and MDRD formula were used to estimate the GFR(eGFR_(CG),eGFR_(MDRD)).The patients were divided into normal GFR groupNGFR,n=589,iGFR≥90 ml·min~(-1)·(1.73 m~2)~(-1)],gently decreased GFR groupGGFR,n=470,60≤iGFR<90 ml·min~(-1)·(1.73 m~2)~(-1)],moderate and sever decreased GFR groupMGFR,n=151,30≤iGFR<60 ml·min~(-1)·(1.73 m~2)~(-1)]based on the K/DOQI suggestion. According to the quartile HbAlc level(7.1%,10.5%),patients were divided into four groups.Patients with HbAlc<7.1% were defined as well-controlled group,with HbAlc≥10.5% as poorly-controlled group.Spearman correlation,t test,Bland-Altman analysis and ROC curve were applied to investigate the bias and accuracy of formula,and the influence of blood glucose on eGFR.Results HbAlc was correlated with iGFR(r=0.17,P<0.01),eGFR_(CG)(r=0.22,P<0.01),eGFR_(MDRD)(r=0.29,P<0.01).The correlation between eGFR_(MDRD) and HbAlc was the strongest.eGFR_(MDRD) seemed to overestimate GFR.However,eGFR_(CG) seemed to underestimate GFR in well-controlled group.Bland-Altman analysis indicated that the bias of eGFR_(MDRD) with iGFR in poorly-controlled group was higher than that in well-controlled group.Compared with poorly-controlled group,the 15% and 30% accuracies of eGFR_(MDRD) in well-controlled group were significantly higher.There was no significant difference between poorly-controlled group and well-controlled group in eGFR_(CG).The bias of eGFR_(CG) with iGFR was significantly higher than that of eGFR_(MDRD) with iGFR.The area under ROC curve was significantly greater in eGFR_(MDRD) than that in eGFR_(CG)(P<0.05)in poorly-controlled group.Conclusions eGFR by MDRD and CG formula in diabetic patients may lead to discrepancies with GFR value,eGFR_(MDRD) overestimates the GFR value.MDRD equation is more accurate and robust in poorly-controlled diabetic patients with mederate renal insufficiency.