修正Gates’法测定肾小球滤过率的初步研究

The modification of the measuring method of glomerular filtration rate with dynamic renal imaging by Gates' method

目的拟合新的肾小球滤过率(GFR)估算公式并验证,从而探讨对Gates’法测定GFR进行修正的可行性。方法收集2011年9月至2016年3月于中山大学附属第三医院就诊的慢性肾脏病(CKD)成人患者资料,符合入选标准病例共229例,以双血浆法测定出GFR(记为rGFR),应用传统Gates’方程计算GFR(记为gGFR),以rGFR为标准将CKD患者按全球肾脏病预后组织(KDIGO)指南分期,将全部病例行分层(按期别)随机抽样分为A、B两组。以A组患者应用肾动态显像得出的双肾放射性摄取率为自变量,以rGFR为因变量行回归分析,拟合出最优曲线,得出新的GFR估算公式。将新公式应用于B组数据,得出新的GFR与rGFR行配对t检验,并运用Spearman秩相关分析,Bland-Altman一致性检验及准确性比较。结果经新拟合方程计算的GFR(线性拟合记为GFRl,曲线拟合记为GFRc)及gGFR均与rGFR密切相关(P<0.01)。GFRl、GFRc与rGFR相关性高于gGFR(r=0.885)。GFRc与rGFR的一致性最好,一致性界限范围最小(-27.8,29.6),偏离度最低[偏差和绝对偏差分别为-0.90、11.64 ml·(min·1.73 m^2)^(-1)]。GFRl次之。Gates’法gGFR与rGFR的一致性最差,一致性界限范围超过事先规定的专业界值60 ml·(min·1.73 m^2)^(-1)。准确性比较,GFRl、GFRc及gGFR的±30%病例符合率分别为75.2%、73.5%、62.8%,gGFR与GFRl、GFRc比较差异具有统计学意义(P<0.01)。结论新拟合方程计算的GFR与双血浆法GFR具有较好的相关性和一致性,其准确性优于Gates'方程,对Gates'方程能够起到一定的修正作用。三次拟合方程评估GFR的一致性及准确度最佳。

Objective A new glomerular filtration rate （GFR） estimation formula is obtained and verified with the purpose of assessing the feasibility to improve the GFR measuring method with Gates＇ method. Method A total of 229 adult patients with chronic kidney disease （CKD） were included From September 2011 to March 2016 in the Third Affitiated Hospital of Sunoyat-sen univesity. The GFR were routinely measured by Gates＇ method （marked as gGFR）, and dual plasma sampling clearance method was used to measure the GFR in these patients （marked as rGFR）. Taking rGFR as the standard, all the patients with CKD were divided into different stages according to the KDIGO guidelines. Patients were stratified through random sampling basing on their stage, being divided into two groups. Taking the bilateral renal uptake rate of patients by the method of dynamic renal imaging in group A as the independent variables, rGFR as the dependent variables to do the regression analysis, so as to find the best-fitting curve, and achieve a new GFR estimate formula. The new formula was applied to group B for estimation of GFR. Compared the result of GFR in group B with rGFR by paired t test, spearman rank correlation analysis, Bland-Altman consistency evaluation and comparing accuracy. Results The new best-fitting curve equations were applied to group B, we got the new data of GFR （the simple equation results marked as GFRl, the results of the cubic equation marked as GFRc）. GFRl, GFRc, gGFR were closely related to rGFR （P〈0.01）. GFRl and GFRc best correlated with rGFR （r=0.885）. Consistency analysis demonstrated that deviation, absolute deviation and limit of agreement of GFRc was lowest among GFRl, GFRc, gGFR and best consist with rGFR, followed by GFRl and gGFR. The limit of agreementof gGFR was out of the pre-set professional cut-off points： 60 ml·（min·1.73 m2）-1. In the term of accuracy analysis, GFRl showed highest concordance with rGFR±30%, and followed by GFRc. The 30% accuracy of GFRl and GFRc were both more than 70% while that of gGFR was less than 70%. Significant differences were also found in GFRl vs. gGFR, and GFRc vs. gGFR, respectively （P〈0.01）. Conclusion The new fitting formulas have better consistency and correlation with dual plasma sampling method, while compared with Gates＇ method, the accuracy of the new fitting formulas is superior to Gates＇ formula. The new fitting formula can improve the performance of Gates＇method.The cubic fitting formula was better potential inpredicting GFR than the linear one.