【作者】 吴杰；

【导师】 陈香美； 谢院生；

【作者基本信息】 中国人民解放军军医进修学院 ， 内科学， 2009， 博士

【摘要】 背景与目的:IgA肾病（IgA nephropathy,IgAN）是世界范围内最常见的原发性肾小球疾病,大多数患者进行性发展,约15-40%的患者在发病10-20年后进展到终末期肾病（ESRD）。肾小管间质损伤（tubulointerstitial lesions,TIL）是影响IgAN预后的重要危险因素。肾活检是目前诊断IgAN、判别其TIL的重要手段。由于肾活检的有创性和风险性,难以反复施行,使许多患者延误了诊断和及时治疗,也不能有效随访病情及观察疗效,因此临床上迫切需要建立无创性诊断手段及寻找无创性标志物。本研究1.通过对IgAN患者的临床病理资料多因素logistic回归分析,筛选出与TIL相关的临床独立影响因素,在此基础上建立判别TIL程度的函数模型,以期为临床无创性评估TIL提供依据。2.对尿中性粒细胞明胶酶相关脂质运载蛋白（NGA1）在IgAN中与临床病理指标的相关性进行分析,探讨其在IgAN患者TIL评估中的意义。3.采用磁珠分离系统联合基质辅助激光解析电离-飞行时间质谱（MALDI-TOF MS）探索IgAN无创性诊断模型及潜在的标志物。对象与方法:1.分析我院肾活检确诊的原发性IgAN患者1493例临床病理资料,在模型组筛选出与TIL相关的临床独立影响因素,建立判别TIL的函数模型,并在独立的验证组中检验其判别效率。2.应用酶联免疫法（ELISA）检测IgAN患者115例,非IgAN肾小球疾病对照组30例,健康对照组30例的尿NGAL水平,分析尿NGAL水平与IgAN患者临床及病理指标相关性。3.利用磁珠分离系统联合MALDI-TOF MS分析IgAN患者32例、非IgAN疾病对照组36例、健康对照组30例的尿液多肽谱,采用ClinProTools2.0软件主成分分析法和SVM算法建立鉴别诊断模型,对差异肽段经混合型串联傅立叶变换离子回旋共振质谱（nano-LC-Q-FT-ICR-MS）进行鉴定。结果:1.（1）1493例lgAN患者中,TIL无病变组12.8%;轻度组47.6%;中度组23.3%;重度组16.4%。（2）TIL与临床指标中年龄、血压、血肌酐、血尿酸、甘油三酯、胆固醇、24小时尿蛋白定量、尿NAG酶呈正相关,与血红蛋白、血白蛋白、尿渗透压呈负相关;与肾活检病理指标中球性硬化、节段损伤、新月体、系膜增殖、血管壁增厚、血管玻璃样变呈正相关。（3）建立函数模型1:在模型组经logistic回归分析,筛选出与有无TIL相关的临床独立影响因素,并建立判别有无TIL的函数模型,（?）_TIL1=e^a/（1+e^a）,其中a=-4.0585+0.9335BP-0.0169Hb+1.4382Scr+0.0026Ua+0.0841ALB+0.2972UP+0.034Age。模型的ROC曲线下面积为0.766,优于各构成指标的ROC曲线下面积,选定界值为0.83,判别函数的灵敏度81.5%,特异度33.3%,准确率88.4%,以此模型在验证组进行验证,验证组的ROC曲线下面积为0.789,灵敏度81.0%,特异度31.2%,准确率87.6%。（4）建立函数模型2:在模型组经logistic回归分析,筛选出与TIL是否达中重度相关的临床独立影响因素,并建立判别TIL程度的函数模型,（?）_TIL2=e^b/（1+e^b）,其中b=-2.0187+0.8482BP-0.0317Hb+1.9032Scr+0.0050Ua+0.2539UP。模型的ROC曲线下面积为0.855,优于各构成指标的ROC曲线下面积。选定界值为0.43,函数判别的灵敏度66%,特异度91.1%,准确率81%,以此模型在验证组进行验证,验证组的ROC曲线下面积为0.851,灵敏度63.6%,特异度91.0%,准确率80.0%。2.（1） IgAN组、非IgAN疾病对照组较健康对照组尿NGAL水平均明显升高,与健康对照组相比有统计学意义,P＜0.01。（2） IgAN组尿NGAL水平与多项临床病理指标明显相关。（3）影响NGAL水平的独立影响因素有血白蛋白水平,尿蛋白定量、尿NAG酶、尿渗透压,以及球性硬化、间质炎细胞浸润、间质纤维化、肾小管萎缩。（4）以尿NGAL水平超过正常对照组2个标准差视为显著增高,联合NAG酶对有无TIL进行评估,ROC曲线下面积为0.850,明显优于血肌酐的0.675,灵敏度83.3%,特异度73.3%,准确率81.2%。3.（1）采用IMAC-Cu²⁺磁珠分离系统联合MALDI-TOF MS提取多肽效果好,随机选取其中10个峰进行观察,日间-日内变异系数均在10%以内。（2） IgAN患者与健康对照、非IgAN疾病对照组的尿液多肽峰度表达存在明显差异,依据肽峰度表达的差异建立鉴别诊断模型,在两两比较中获得的模型鉴别IgAN与健康对照时灵敏度、特异度均为100%;鉴别IgAN与疾病对照时灵敏度为85.7%,特异度为76.5%。三组比较获得的模型在鉴别健康对照组准确率100%,鉴别IgAN组准确率71.4%,鉴别疾病对照组准确率76.9%。（3）鉴定得到了差异肽段m/z1913.14序列图SGSVIDQSRVLNLGPITR,经检索确定为uromodulin（尿调蛋白）的肽段。此肽段表达的峰度差异用于鉴别IgAN与健康对照组时ROC曲线下面积为0.998;在鉴别IgAN与疾病对照组时ROC曲线下面积为0.815。结论:1.IgAN患者中TIL的发生率较高,TIL与多项临床病理指标相关。通过对IgAN患者TIL发生的临床独立影响因素分析建立了有无TIL及TIL是否达中重度的判断函数模型,模型可重复性好,且有较好的ROC曲线下面积,对TIL病情判断的准确性分别为88.4%、81%。为临床上无创性判断IgAN患者TIL的程度、判断病情、随访疗效提供了参考。2.IgAN患者尿NGAL水平随临床病情的加重及病理指标积分的增加而明显升高,可以反映IgAN病情程度。在联合尿NGAL与尿NAG酶判别有无TIL时明显优于血肌酐及单纯的尿NGAL、尿NAG酶、尿渗透压等指标。3.采用IMAC-Cu²⁺磁珠分离系统联合MALDI-TOF MS分析人尿液多肽谱有较好的稳定性及可重复性。通过尿肽的峰度差异建立了IgAN与健康对照、疾病对照的鉴别诊断模型,有较好的灵敏度及特异度。m/z1913.14为uromodulin（尿调蛋白）的肽段,其特殊的强度范围可以作为诊断IgAN的一个可参考的无创性指标。更多还原

【Abstract】 Background and Objective:IgA nephropathy（IgAN） is the most common glomerulonephritis worldwide,which develops progressively in most patients with IgAN and 15-40%of of which eventually develops into end stage renal disease（ESRD） in 10-20 years.Tubulointerstitial lesion（TIL） is one of the strongest predictors of IgAN outcome.Up to now,renal biopsy has been the important method for diagnosing IgAN and evaluating assessing TIL.Due to the facts that renal biopsy is an invasive procedure with inherent risks,and difficult to be performed repeatedly,many patients can neither be diagnosed and treated in time,nor be followed up effectively.Therefore,it is necessary to find non-invasive diagnostic methods and markers of IgAN and TIL.In this study, through multi-factor logistic regression analysis of clinical and pathological data of patients with IgAN,clinical independent affecting factors were screened out associated with TIL,and function models were established for evaluating the degree of TIL.The association between urine neutrophil gelatinase-associated lipocalin（NGAl） and clinicopathological parameters in IgAN were also analyzed to explore the significance of NGAl in assessing TIL of IgAN.Besides,the magnetic separation system combining matrix-assisted laser desorption ionization -time of flight mass spectrometry（MALDI-TOF MS） was applied to explore non-invasive diagnosis model and potential markers of IgAN.Subjects and Methods:1.Clinical and pathological data of 1493 IgAN cases were analyzed,respectively.Clinical independent affecting factors associated with TIL were screened out from the model group to set up functional model assessing TIL,which was verified in the verifying group.2.Urine NGAL level was detected by ELISA in 115 cases with IgAN,30 cases with non-IgAN glomeruionephritis, and 30 healthy control persons in order to analyze the association between urinary NGAL level and clinicopathophysiological parameters.3.The magnetic separation system combining MALDI-TOF-MS was used for analyzing urinary peptide spectra of 32 patients with IgAN,36 patients with non-IgAN glomerulonephritis,and 30 healthy controls.ClinProtools 2.0 software principal constituent analysis method and SVM algrithm were applied for establishing diagnostic model.And differential peptides were discriminated with nano-LC-Q-FT-ICR-MS.Results:1.（1） The incidence of TIL was 87.2%in 1493 IgAN patients,with percentages of slight/moderate/severe TIL were 47.6%,23.8%,and 16.4%. （2）The degrees of TIL were positively correlated with age,blood pressure,serum creatinine,blood uric acid,triglyceride,cholesterol,urine protein,and NAG enzyme.The degrees of TIL were negatively correlated with haemoglobin,blood albumin,urine osmotic pressure;positively correlated to glomerulosclerosis, segmental damage,crescent,mesangial proliferation,arterial wall thickening,and arterial hyaline change.（3） The first functional model:Through logistic regression analysis in the model group,clinical independent affecting factors were screened out associated with or without TIL,and function models were established for confirming TIL,（?）_TIL1=e^a/（1+e^a）,a=-4.0585+0.9335BP-0.0169Hb+1.4382Scr+ 0.0026Ua+0.0841ALB+0.2972UP+0.034Age.AUC of the ROC was 0.771. Using the cutoff of 0.83,the sensitivity,specificity and the accuracy was 81.5%, 33.3%and 88.4%,respectively.（4） The second functional model:Through logistic regression analysis in the model group,clinical independent affecting factors were screened out associated with or without moderate-to-severe TIL,and function models were established for assessing TIL degrees,（?）_TIL2=e^b/（1+e^b）,b= -2.0187+0.8482BP-0.0317Hb+1.9032Scr+0.0050Ua+0.2539UP.AUC of the ROC was 0.855.Using the cutoff of 0.43,the sensitivity,specificity,and the accuracy was 66%,91.1%,and 81%,respectively.The diagnostic efficacy in the estimation group was similar to that of verifying group.2.（1）The level of urine NGAL in IgAN or non-IgAN glomerulonephritis patients was higher than that in healthy controls（P＜0.001）.（2） The level of urine NGAL in IgAN was significantly correlated with many clinical and pathology parameters. （3） The independent influencing factors on urine NGAL included blood albumin, urine protein,NAG enzyme,urine osmotic pressure,and tubulointerstitial lesions. （4） To evaluate whether or not TIL existed,ROC analysis of urine NGAL combining NAG enzyme showed that AUC was 0.850,while the AUC for serum creatinine was 0.675.3.（1）The whole system of MB-IMAC-Cu²⁺ combining MALDI-TOF MS was effective in isolation of urine peptides,and the CV was less than 10%.（2） There were significant differences in the urine peptides peaks among IgAN group, non-IgAN glomerulonephritis,and healthy controls.The most optimal classification with highest accuracy eventually was established for the discrimination of IgAN patients from healthy individuals and other glomerulonephritis.When it was evaluated by cross-validation,the sensitivity and specificity to distinguish IgAN versus healthy controls were both 100%,and sensitivity and specificity to distinguish IgAN versus other glomerular diseases were 85.7%and 76.5%.（3） The peptide with m/z 1913.14 was identified as fragments of uromodulin.ROC analysis for m/z 1913.14 to distinguish IgAN versus healthy controls showed that the AUC was 0.998.ROC analysis for m/z 1913.14 to distinguish IgAN versus other glomerular diseases showed that the AUC was 0.815.Conclusions:1.The incidence of TIL in IgAN was high.The degrees of TIL were obviously correlated with many clinical and pathology parameters.The functional models for assessment of TIL were established with good sensitive,accuracy and good reproducibility,which can help predicting the degree of TIL and providing intervention early.2.The levels of urine NGAL in IgAN were obviously correlated with many clinical and pathology parameters,and could reflect the degree of illness.Combination of urine NGAL and NAG enzyme was better in evaluating whether or not TIL existed than serum creatinine,urine NGAL,urine NAG,or osmotic pressure alone.3.Analysis of urine peptides patterns by MB and MALDI-TOF MS was a non-invasive diagnostic tool with good sensitivity and specificity.Urine peptide marker of m/z 1913.14 was identified as the fragment of uromodulin,which may be used as a non-invasive way for clinical diagnosis of IgAN.更多还原