节点文献
高发区人群食管、贲门癌早诊早治研究
Study on Early Diagnosis and Treatment of Esophageal and Cardia Cancer among Population in High Incidence Areas
【作者】 张立玮;
【导师】 王士杰;
【作者基本信息】 河北医科大学 , 外科学, 2007, 博士
【摘要】 目的:食管癌是世界上最常见恶性肿瘤之一,显著的地域性分布是其流行病学突出特征,高、低发区发病率和死亡率相差可达500倍。我国太行山南麓是世界上主要的食管癌高发区,近年来研究发现在这一地区贲门癌的发病率也处于较高水平,存在食管癌和贲门癌共同高发的现象。由于早期食管、贲门癌发病隐匿,不易发现,高发区就诊病例多为中晚期患者,术后5年生存率不足25%,花费巨大但治疗效果较差。因此,若要改变这种以治疗中晚期肿瘤为主的模式,降低食管、贲门癌的发病率、提高患者生存率及生存质量,早期发现、早期诊断、早期治疗是现阶段切实有效的防治途径。目前,内镜检查结合病理组织学诊断是发现早期癌和癌前病变最有效的手段和方法,而探讨遗传易感性与环境危险因素之间的相互作用,筛选和建立用于早期诊断的生物学诊断指标是当前肿瘤防治研究的热点问题。本研究选取太行山区上消化道癌高发现场,选定目标人群,在流行病学调查的基础上建立筛查研究队列和肿瘤防治数据库,对高危人群进行有计划内镜筛查,对发现的早期癌及癌前病变积极干预的同时,深入观察癌前病变微细形态,探讨食管、贲门癌发病危险因素,筛选和建立用于高危人群预警的生物学标志,以期为食管贲门癌高危人群监测、早期诊断和治疗奠定重要基础。研究内容如下:①对流行病学调查和内镜筛查结果进行综合分析,初步探讨碘染色对早期食管癌及癌前病变诊断的敏感度和特异度,明确贲门部病变分布情况,了解相关危险因素与各级病变检出率间的关系,为高发区肿瘤预防的开展提供线索;②通过部分内镜初筛个体间隔较短时间复查结果,分析食管及贲门癌前病变个体的滞留时间及自然演变进程,探讨间期癌发生的可能原因,寻找最佳的内镜筛查方案;③通过对比涉县食管癌及贲门癌发病率、死亡率情况,分析贲门癌流行强度及内镜筛查结果,并应用放大内镜和扫描电镜,观察贲门病变局部粘膜微细形态结构,分析粘膜腺管开口类型与病理组织改变的关系,探讨放大内镜和扫描电镜辅助诊断贲门病变的可能性,以期提高内镜对贲门癌前病变的识别和检出,同时对检出的早期贲门癌及癌前病变患者,采用EMR和APC方法进行内镜微创治疗,随访观察治疗效果及尚存问题;④对食管、贲门癌患者进行炎症因子IL-8的SNP基因分型,初步明确高发区个体遗传易感性与环境危险因素在肿瘤发生中的交互作用,试图界定具有某易感基因型并暴露于相关危险因素的高发区个体,作为肿瘤高危人群进行重点预防,并利用已知的保护措施对其进行干预,追踪观察;⑤为寻找简便、易行,敏感、特异的分子标志物用于无症状人群的初筛,应用SELDI-TOF-MS方法和蛋白芯片系统筛选并建立食管、贲门癌及癌前病变患者血清蛋白指纹图谱模型,浓聚高危人群,使高发区大规模人群内镜筛查更具有针对性。方法第一部分:食管癌高发区高危人群内镜筛查与肿瘤相关危险因素分析选择食管癌发病率较高的丘陵地区河北省磁县固义乡21个自然村为筛查社区,以40-69岁人群为研究对象,建立队列并进行流行病学、危险因素基线调查,实行计算机数据库管理。于2005到2006两年间,采用直接电子胃镜检查方法共筛查3660例。全部病例均行食管、贲门、胃及十二指肠球部检查,全食管粘膜常规1.2%-1.5%碘液染色,将粘膜不染或淡染视为染色阳性,活检行病理组织学检查,食管多灶病变以病理诊断最重计入统计,对肉眼可疑但染色阴性病灶也应活检。另对贲门部病变详细记录其位点分布并取活检。全部标本双盲法病理组织学确诊。统计内镜筛查结果,分析内镜筛查率,判断碘染色对早期癌及癌前病变诊断的灵敏度和特异度,明确贲门部病变分布情况,探讨相关危险因素与食管、贲门各级病变检出率间的关系。第二部分:高发区食管、贲门癌前病变自然史与内镜筛查间隔的研究涉县位于河北省西南端,地处太行山南麓的三省交界处。本研究自2001年3月至2005年6月,在涉县固新、神头两乡的10个自然村对40~69岁居民行内镜筛查1514人,筛查率均达70%以上。对301例(19.9%)接受了首轮内镜筛查者间隔1-50个月时间进行了复查,方法同本研究第一部分,前后对比检出结果以观察进展较快个体癌及癌前病变的滞留时间、了解多点起源和一次性筛查的漏诊情况,用以指导筛查实践。第三部分:食管癌高发区贲门癌流行强度分析与内镜早诊早治研究2000年至2004年在食管癌高发区普查时对涉县贲门癌进行单独登记,统计其发病率与死亡率,并将涉县食管癌、贲门癌和胃癌的两率(发病率和死亡率)与相邻磁县和林县进行比较,分析食管癌高发区贲门癌所占比例,并结合上述地区三种癌及癌前病变的内镜检出数据,判断内镜对早期贲门癌及癌前病变的检出能力。应用Olympus GIF-Q240Z型放大电子胃镜辅以3%乙酸染色观察贲门粘膜微细形态,分析粘膜小凹类型与病理组织学诊断的符合率,同时行扫描电镜观察其超微结构变化,以了解病变发展的形态学变化。采用14C-尿素呼气试验和病理组织学检测贲门部病变H.pylori的感染情况,并分析比较其在12点位和6点位的感染率的高低,探讨贲门各类病变不同位点与H.pylori感染之间的关系。对内镜检出和病理组织学证实的55例早期贲门癌及癌前病变患者,行内镜治疗,治疗方法主要包括EMR和APC两种,治疗后定期复查,追踪随访以评价疗效。第四部分:IL-8基因多态性与食管、贲门癌发病风险的关联研究来自河北高发地域320例食管鳞状细胞癌(Esophagealsquamous carcinoma, ESCC)、340例贲门腺癌(Gastric cardiac adenocarcinoma,GCA)和404例正常对照,均经内镜检查病理组织学确诊及H.pylori检测。收集其个人史、疾病史、上消化道癌家族史,并排除自身存在其它恶性肿瘤史。抽取2ml外周静脉血,经柠檬酸钠抗凝,提取DNA,应用聚合酶链反应-限制性片段长度多态性(Polymerase chain reaction-restriction fragment length polymorphism analysis, PCR-RFLP)方法检测研究对象IL-8各基因型的分布情况,任选显示不同基因型的标本进行DNA测序。比较各基因型频率的观察值与预期值并进行卡方检验行Hardy-Weinberg平衡分析。病例组与对照的基因型分布比较采用行×列表卡方检验。以非条件Logistic回归法计算经年龄、性别校正的相对风险度比值比(Odds ratio,OR)及其95%可信区间(Confidence interval, CI)。第五部分:高发区食管、贲门癌及癌前病变患者血清蛋白指纹图谱检测及其筛查价值研究标本取自2004年9月至2005年12月期间高发区磁县、涉县筛查人群及门诊患者,均经内镜碘染色检查、病理组织学确诊。选取检出的26例食管鳞状上皮轻度不典型增生(Esophageal mild dysplasia, EDYSⅠ)、26例中度不典型增生(Esophageal moderate dysplasia, EDYSⅡ)、11例重度不典型增生(Esophageal severe dysplasia,EDYSⅢ)、15例早期食管癌(Early esophageal cancer,EEC)、36例进展期食管癌(Advanced esophageal carcinoma,AEC)、21例贲门腺上皮轻度不典型增生(Gastric-cardia mild dysplasia,GDYSⅠ)、10例中或重度不典型增生(Gastric-cardia moderate or severe dysplasia,GDYSⅡ~Ⅲ)、34例GCA患者,另随机选取年龄和性别匹配的38例NOR一并纳入本研究。采用CM10蛋白芯片及SELDI-TOF-MS技术检测其血清蛋白质谱,结果采用浙江大学肿瘤研究所蛋白芯片数据分析系统(Zhejiang university cancer insititute Proteinchip data analysis system,ZUCI-PDAS)软件包分析,对每个质荷比峰值做Wilconxon秩和检验,选出p值最小的10个峰进一步分析。将10个峰的任意组合用于支持向量机模型的输入,选出Youden指数最高的组合作为候选标志物,分别建立食管、贲门癌及癌前病变血清蛋白指纹图谱模型,并用留一法交叉验证作为评估模型判别效果的方法。结果第一部分:食管癌高发区高危人群内镜筛查与肿瘤相关危险因素分析1直接内镜筛查3660人,筛查率达71%。食管轻、中、重度不典型增生、早期癌和浸润性癌的检出率分别为11.9%、2.1%、0.4%、0.8%和0.3%;贲门轻、中、重度不典型增生、早期癌和浸润性癌的检出率分别为1.1%、0.4%、0.2%、0.2%和0.2%。2无论食管和贲门,各级不典型增生及癌的检出率都随年龄增加呈显著上升趋势(食管x2 =130.9,P=0.00;贲门x2 =40.0,P=0.00)。3早期食管癌100%(31/31)碘染阳性,重度不典型增生93.8%(15/16)碘染色阳性,正常及炎症组食管粘膜19.9%碘染色阳性。前两者合并与正常及炎症组碘染色阳性比率差异有显著统计学意义(χ~2=169.8,P<0.01)。4内镜检出贲门癌前病变62例,病变位于12点~3点位占69.4%(43/62),其他点位占30.6%(19/62),点位分布差异有显著统计学意义(P<0.01)。早期癌检出8例,位于12点~3点位占87.5%,其他点位占12.5%(1/8),点位分布差异有显著统计学意义(P<0.01)。5食管癌及癌前病变在男性、上消化道癌家族史阳性、吸烟者中的检出率总体上分别高于女性、家族史阴性、不吸烟者(χ~2值分别为17.0、22.7、15.8,P值分别为0.005、0.000、0.007),但饮酒史与检出率无关(χ~2 =4.5,P=0.5)。6贲门癌及癌前病变在男性和上消化道癌家族史阳性中检出率总体上分别高于女性和家族史阴性者(χ~2值分别为34.6和14.3, P值分别为0.000和0.004),吸烟和饮酒史与检出率无关(χ~2分别为8.2和7.5,P值分别为0.16和0.19)。第二部分:高发区食管、贲门癌前病变自然史与内镜筛查间隔的研究1内镜初筛1514例,间隔1-50个月复查301例,其中进展为重度不典型增生、原位癌、粘膜内癌及浸润性癌共16例。2重度不典型增生(7例)中:1例首检为正常,间隔13个月检出重度不典型增生,1例首检为基底细胞增生,间隔7个月检出,4例首检为轻度不典型增生,分别间隔3个月、4个月、4个月、10.5个月检出,1例首检为中度不典型增生,间隔12.5个月检出。3原位癌及粘膜内癌(6例)中:1例首检为轻度不典型增生,间隔48个月检出,2例中度不典型增生分别间隔4个月、13个月检出2例,3例重度不典型增生分别间隔3.5个月、9个月、17.5个月检出。4浸润性癌(3例)中:1例中度不典型增生,间隔50个月检出,2例重度不典型增生间隔14个月和19个月检出。第三部分:食管癌高发区贲门癌流行强度分析与内镜早诊早治研究1贲门癌流行病学调查1.1涉县贲门癌的发病率和死亡率涉县2000-2004年贲门癌发病率男女性分别为69.9/10万和41.5/10万,死亡率为54.3/10万和33.2/10万。与食管癌、胃癌合计占全身恶性肿瘤死亡率的78.7%和71.9%,发病率的79.8%和68.6%。1.2涉县上消化道癌及癌前病变内镜检出情况比较涉县、林县和磁县应用内镜碘染色指示性活检的方法对40-69岁人群筛查,贲门粘膜内癌检出率分别为0.6%、0.7%和0.1%。其中涉县、林县贲门粘膜内癌和食管原位癌检出率的比例约1:4;涉县、磁县贲门腺上皮和食管鳞状上皮重度不典型增生检出率的比例也约为1:4;轻、中度不典型增生的检出率贲门也远低于食管。2贲门部病变粘膜微细形态、超微结构观察及H.pylori感染情况2.1放大内镜观察贲门部病变粘膜微细形态将A、B、C、D四种粘膜小凹类型和贲门浅表性炎、活动性炎、萎缩性炎进行相关分析,结果表明上述四种小凹类型与慢性炎症严重程度之间关系密切。本组病例中肠上皮化生的粘膜小凹类型可表现为B、C、D型,其中C型所占比率最高66.13%(41/62),与B、D型比较统计学有显著性差异。本组不典型增生病变共23例, C型占48.0%(11/23), D型占21.7%(5/23),E型占30.4%(7/23),其中E型在重度不典型增生中所占比例达64.28%(5/7),明显高于轻度及中度不典型增生(P<0.05)。早期贲门癌粘膜小凹均为E型改变,为100%(11/11)。轻、中、重度不典型增生三组病变与小凹类型变化进行列关联检验,χ~2=10.31,pearson系数=0.644,P值<0.05。2.2扫描电镜观察贲门病变粘膜小凹的超微结构改变扫描电镜的低倍镜观察粘膜小凹随着分型的变化,逐渐延长和不整,从较规则的点状、棒状到树枝、斑块状直至不规则状甚至结构消失。扫描电镜高倍观察:A型粘膜小凹呈规则的圆形,由2-4个细胞相互围成,细胞形态规整,排列整齐,表面附微绒毛;B型粘膜小凹呈短棒状,围成粘膜小凹的细胞数目增多,细胞形态及排列变化不大,开口变形;C型粘膜小凹呈树枝状,围成粘膜小凹的细胞数目进一步增多,使粘膜小凹弯曲变形呈长裂隙状,由中心至四周腔隙逐渐变细,细胞排列不整;D型粘膜小凹呈斑块状,围成粘膜小凹的细胞排列紊乱,可见少量细胞变性坏死;E型粘膜小凹结构破坏,细胞变性坏死明显,细胞表面微绒毛消失。断面图像显示腺管壁柱状上皮细胞呈砖墙样排列,细胞易脱落。2.3贲门病变H.pylori感染情况贲门癌及癌前病变与H.pylori感染关系密切(χ~2=32.35,P值<0.05),各级病变中以活动性贲门炎的H.pylori感染率(81.8%)最高。全部病例均分别在贲门12、6点位检测H.pylori感染情况。结果显示,12点阳性在全部阳性中所占比率为96.48%,6点阳性比率为65.10%;两位点H.pylori阳性率比较差异有统计学意义(χ~2=108.18,P值<0.05)。3早期贲门癌及癌前病变的内镜治疗结果本组55例患者中,病灶位于贲门脊根部即小弯偏后壁处(12点至3点位)43例占78.18%;病灶位于其他部位(前壁、大弯侧)12例占21.82%,组间比较差异有统计学意义。15例行EMR治疗患者病灶完全切除者11例占73.3%,其余4例残余病灶术中行APC烧灼治疗。40例APC治疗病例首次治疗均覆盖病灶全部范围,早期癌平均治疗次数为2.9次,癌前病变为1.9次,治疗成功率分别为87.5%和100%。EMR治疗前后病理比较发现,术前4例中度不典型增生术后证实1例为粘膜内癌;7例重度不典型增生术后证实2例为粘膜内癌。全部治疗病例中9例发生出血,无穿孔等其他并发症的发生。按治疗术后1个月、3个月、6个月、12个月,以后每年内镜复查一次随访。第四部分:IL-8基因多态性与食管、贲门癌发病风险的关联研究1 PCR-RFLP分型显示,贲门癌组A等位基因频率(49.6%)明显高于健康对照组(43.1%),χ~2=6.260,P值为0.012,差别有统计学意义,而在食管癌组各等位基因出现的频率未见显著性差异。GCA和健康对照组基因型频率分布差异有统计学意义(χ~2=6.299 ,P=0.043)。ESCC和健康对照组相比未见统计学差异。2 IL-8基因型与食管、贲门癌危险度的比较:纯合突变基因型AA在GCA组中出现频率(24.7%)显著高于健康对照组(18.3%) ,经性别、年龄、吸烟、上消化道癌家族史和H.pylori感染因素校正后OR=2.014,95%CI=1.017-3.990。在ESCC组中AA基因型出现的频率(15.6%)比正常人(18.3%)也有降低趋势(OR=1.901, 95%CI=0.952-3.794),但组间差异无统计学意义。3 IL-8基因型分布与食管、贲门癌危险因素的交互作用3.1吸烟阳性组中未发现各基因型的改变与ESCC、GCA的发病风险有关,非吸烟组健康对照组的基因型频率与ESCC、GCA组相比差异也无显著性(P>0.05);3.2 UGIC阳性组,ESCC(23.1%)和GCA组(19.8%)的AA基因型频率明显高于健康对照组(13.8%),P值均<0.05。以纯合野生基因型TT为参照,AA基因型UGIC家族史阳性个体ESCC、GCA的发病风险显著升高,经性别、年龄、吸烟、H.pylori(GCA组)校正后的OR值分别为2.378,(95%CI=1.075-5.258),14.895(95%CI=2.889-76.516)。而在UGIC阴性组,各基因型频率未发现有统计学意义;3.3 H. pylori感染阳性组,AA基因型频率明显高于健康对照组,P=0.017,以TT基因型为参照,AA基因型H. pylori感染阳性个体GCA的发病风险显著升高,经性别、年龄、吸烟、家族史校正后的OR值为3.520(95%CI=1.249-9.918),而在H. pylori感染阴性组中三种基因型频率未见统计学差异。第五部分:高发区食管、贲门癌及癌前病变患者血清蛋白指纹图谱检测及其筛查价值1食管病变血清蛋白指纹图谱模型的建立1.1应用SELDI-TOF-MS技术结合支持向量机方法分别建立食管癌及癌前病变血清蛋白指纹图谱模型共九个,即诊断模型EDYSⅠvs NOR、EDYSⅡvs NOR、EDYSⅢvs NOR、EEC vs NOR、AEC vs NOR、EDYSⅠvs AEC、EDYSⅡvs AEC、EDYSⅢvs AEC、EEC vs AEC。1.2诊断模型EDYSⅠvs NOR、EDYSⅡvs NOR、EDYSⅢvs NOR、EEC vs NOR、AEC vs NOR的特异性(Specificity,Spe)分别为92.11%、84.21%、81.58%、81.58%、89.47%,敏感性(Sensitivity,Sen)分别为42.31%、73.08%、45.45%、80.00%、83.33%。1.3诊断模型EDYSⅠvs AEC、EDYSⅡvs AEC、EDYSⅢvs AEC、EEC vs AEC的Spe分别为92.31%、80.77%、90.91%、73.33%,Sen分别为80.56%、83.33%、94.44%、91.67%。1.4在上述诊断模型中,质荷比峰Mass/Charge(M/Z)值4291、4975、5644、5664、8775Da重复出现。2贲门病变血清蛋白指纹图谱模型的建立2.1应用上述技术分别建立贲门癌及癌前病变血清蛋白指纹图谱模型共六个,即诊断模型GDYSⅠvs NOR、GDYSⅡ~Ⅲvs NOR、GCA vs NOR、GDYSⅠvs GCA、GDYSⅡ~Ⅲvs GCA、GDYSⅠvs GDYSⅡ~Ⅲ。2.2诊断模型GDYSⅠvs NOR、GDYSⅡ~Ⅲvs NOR、GCA vs NOR的Spe分别为86.84%、100%、94.74%,Sen分别为61.90%、90%、88.24%。2.3诊断模型GDYSⅠvs GCA、GDYSⅡ~Ⅲvs GCA的Spe分别为90.48%、80.00%,Sen分别为88.24%、91.18%。2.4诊断模型GDYSⅠvs GDYSⅡ~Ⅲ的Spe为100%,Sen为80.00%。2.5在上述诊断模型中,质荷比峰M/Z值3271、6891、8571Da重复出现。结论第一部分:食管癌高发区高危人群内镜筛查与肿瘤相关危险因素分析1在高发区大人群中,以直接内镜检查结合碘染色指示性活检技术作为筛查手段,确能有效地检出早期食管癌和癌前病变。2早期贲门癌及癌前病变在12点~3点位的检出率,与其他位点检出率比较差异有显著性,提示在高发区筛查中重视对上述高发位点的观察和活检,可提高癌前病变的检出。3食管癌及癌前病变在男性、上消化道癌家族史阳性、吸烟者中的检出率高于女性、家族史阴性、不吸烟者,但饮酒史与检出率无关。贲门癌及癌前病变在男性和上消化道癌家族史阳性中检出率高于女性和家族史阴性者,吸烟和饮酒史与检出率无关。4提高高发区人群对内镜筛查的认知度和依从性,采取切实措施动员相对高危个体进入筛查,可进一步提高早期癌及癌前病变的检出。第二部分:高发区食管、贲门癌自然史与内镜筛查间隔的研究1大人群内镜筛查后纵向监测观察发现间期癌及癌前病变的发生,分析原因有三:①癌前病变的滞留时间个体差异比较大,部分病例恶性程度较高、进展较快;②多点起源造成的同时或异时性异位再发的;③首次筛查的漏诊。2高危人群筛查应避免一次性大范围筛查多见,后续复查跟不上的做法,尤其应适当缩短癌前病变的筛查间隔,克服一次性筛查的漏诊和片面性。3本组建议初筛为炎症、基底细胞增生和轻度不典型增生间隔二至三年复查一次,中度不典型增生及以上个体应半年至一年内复查一次。这一方案有待于在深入研究食管、贲门癌前病变自然史与临床发癌之间关系的基础上,进一步修改和完善。第三部分:食管癌高发区贲门癌流行强度分析与内镜早诊早治研究1太行山南麓食管癌高发区存在贲门癌共同高发的现象,重视对贲门癌的防治研究是整体上降低该地区上消化道癌两率的关键。早期贲门癌及癌前病变检出率相对较低,提高内镜检出和识别能力是亟待解决的问题。2放大内镜对贲门病变粘膜小凹类型的判定与病理组织学改变密切相关。依据粘膜小凹类型指导活检,有助于提高病变诊断的准确性。C型粘膜小凹是肠上皮化生的特征性表现。小凹类型由C→E型显示不典型增生程度逐渐加重,E型为重度不典型增生和早期癌的特征性表现。3贲门病变与H.pylori感染密切相关,以活动性贲门炎的H.pylori感染率最高,提示H.pylori感染可能是贲门炎的危险因素。其中12点位阳性率高于6点位,提示了贲门癌高发位点H.pylori高感染情况的存在。4扫描电镜清晰地显示从正常到早期贲门癌各阶段的粘膜小凹超微结构,对研究放大内镜观察到的粘膜小凹类型变化形成机理具有指导意义。5 EMR和APC治疗早期贲门癌及癌前病变安全有效,在高发区是值得推广的内镜微创治疗方法。应将中度不典型增生患者纳入治疗范围。粘膜下注药可做为治疗方式选择的方法之一。第四部分:IL-8基因多态性与食管、贲门癌发病风险的关联研究1携带IL-8-251AA基因型个体显著增加贲门癌发病风险,食管癌患者发病风险与IL-8-251位点单核苷酸多态无相关性。2上消化道癌家族史阳性是食管、贲门癌的易感因素,家族史阳性携带IL-8-251AA基因型的个体贲门癌的发病风险显著增高,食管癌患者虽未提示基因型与之发病有关,但有遗传家族史背景的个体AA基因型成为易感因素。3 H.pylori感染是贲门癌发病的危险因素,H.pylori感染阳性携带IL-8-251AA基因型的个体显著增加贲门癌的发病风险。4吸烟是食管癌发生的危险因素,与贲门癌的发病风险不明显,尚未发现IL-8各基因型与吸烟状况存在交互作用。第五部分:高发区食管、贲门癌及癌前病变患者血清蛋白指纹图谱检测及其筛查价值1.本研究利用SELDI-TOF-MS技术检测食管、贲门癌及癌前病变患者的血清蛋白质谱变化,结合支持向量机算法共建立十五个蛋白指纹图谱模型,其中十个模型的灵敏性和特异性双高。2.进一步对比分析发现,重复出现的五个质荷比峰(M/Z值为4291、4975、5644、5664、8775Da)对食管病变有相似分类作用;三个质荷比峰(M/Z值3271、6891、8571Da)对贲门病变有相似分类作用,可能是与食管、贲门癌变过程密切相关的特异性肿瘤标志物。3.本课题采用的支持向量机(SVM)算法是一种新分类技术,相对于常用的决策树和人工神经网络算法,优点在于实现了统计学习理论中的结构风险最小化,从而具有较好泛化能力,确保数据分析和模型建立的准确性和有效性。4.上述蛋白指纹图谱模型,能良好区分食管、贲门癌及癌前病变和正常对照,应进一步扩大样本量验证其重复性和有效性;对实验中检测到的大量有差异质荷比峰,如能开展分离、纯化、鉴定等后续工作,对高发区大人群的初筛具有广阔的应用前景。
【Abstract】 Objective: Esophageal cancer is one of the most common malignant tumors in the world. A very remarkable feature of the cancer is the highly centralized occurrence in the high risk regions, with the difference in incidence and mortality rates as large as 500 times exist between the high and low risk areas. The South Taihang mountain region of our country has been well recognized as the highest incidence area for esophageal cancer in the world. Recently it is know that this area not only bears high risk for esophageal cancer, but also for cardia cancer. However, early diagnosis of the cancer is hard to achieve because patients in the early stage have no obvious symptoms, and majority of patients treated in tumor hospital are in the late stage, so the five year survival rates after operation were no more than 25 percent despite the huge cost . So early diagnosis and treatment is the key to improve the survival. Presently, endoscopy with bioposy is the most effective screen method to discover early carcinoma and precancerous lesion. The present study performed a populational screen in the high incidence county of Cixian, and also evaluated the value of some potential biomarkers for early diagnosis.Our study selected the high incidence region of Cixian and Shexian counties in the area, established cohort and data bank through epidemiological investigation, then endoscopically screened the 40 to 69 age groups and treated early carcinoma and precancerous lesion with mucosal resection. At the same time, observed and compared the detection rates of precancerous lesions, and investigated the value of biomarkers for early diagnosis. The detailed are as followings:①In our study, we analyzed endoscopic screening rates and epidemiology data, calculated specificity and sensitivity of iodine staining under endoscopy to diagnoze carcinoma and precancerous lesions, identified distribution of cardia lesions, investigated the correlation between the screening rate and high risk factors. The above measure can provide help for prevention and cure in high risk region;②Analyzed the results of periodic screening with shorter intervals, investigate sojourn time and natural history of precancerous lesions, to look for the optimal screening schema;③In order to improve the diagnosis rate of cardiac diseases , we estimated the epidemic strength of esophageal and cardiac cancer in the high incidence of Shexian, and watched the microstructures of mucosa with different cardiac diseases using magnifying endoscope and scanning electron microscope, evaluated the diagnostic value,and at the same time treat the early cardia cancer and precancerous lesion by EMR or APC;④Investigated the association of interleukin(IL)-8 single nucleotide polymorphism(SNP) with the susceptibility to esophageal and cardia cancer to clarify the relationship of genetic predisposition and environmental factor with the risk of tumor, search the high risk individuals and defend them;⑤In order to improve the efficency of endoscopic screening, we established finger printing model of serum protein by SELDI-TOF-MS and array to look for more convenient, sensitive and specific early biomarker for preliminary screening of the symptomless people in the high incidence areas.MethodsPart 1 Endoscopic screening and analysis of risk factors in high incidence area of esophageal carcinomaFrom 2005 through 2006, our study selected 21 villages with higher incidence rate in Guyi community, established cohort and data bank through epidemiological investigation, then endoscopically screened the 40 to 69 age group. 3660 cases were examinated from esophagus to duodenal ampulla by using electronic gastroscope with mucosal iodine staining. All visible understained and unstained lesions were judged positive. And then, biopsy were obtained from above-mentioned and suspicious negative areas. Bioposying from the cardia were performed with the exact clock number recorded. All the histological diagnoses were double-blindly made. In our study, we analyzed endoscopic screening rates, calculated specificity and sensitivity of iodine staining under endoscopy to diagnoze carcinoma and precancerous lesions, identified distribution of cardia lesions, investigated the correlation between the screening rate and high risk factors. Part 2 The intermittent time of Endoscopic screening in high incidence area and the natural history of precancerous lesions from esophagus and cardiaShexian county locates southwest of hebei province. It is junction of three province in south taihang mountain. Endoscopic screening was carried out among the 40 to 69 age group in 10 villages of Guyi and Shentou community. 1514 cases residents were examined by using endoscopy. The screening rate was more than 70%. Periodic endoscopic screening was performed among 301 residents within one to fifty months after the first screening. The method was same as above. We compared the periodic screening results, then investigated the sojourn time of rapid developing esophageal and cardiac precancerous lesions, multifocal carcinogenesis, and false negative results, in order to guide the screening.Part 3 Epidemic strength, early diagnose and treatment of cardia cancer in high incidence region of esophageal cancerIn this part, firstly, We estimated the epidemic strength of cardiac cancer in the high incidence region of Shexian county from 2000 to 2004, through comparing the morbidity and mortality of esophageal, cardia and stomach cancer in Shexian county with in Cixian and Linxian county. The detection rates of endoscopy were estimated in cardia cancer and its precancerous lesions. Secondly, we watched the microstructures of mucosa with different cardia lesions by using magnifying endoscope with 3 percent acetic acid spraying, and analysed the coincidence of mucosa pit patterns and histophathological results. At the same time, observed the changes of mucosa ultrastructure with scanning electron microscope. We also detected the infection of H.pylori in cardia lesions by 14C-urea breath test and histopathology , analysed the infection rate of 12 and 6 clock bit repectively, to estimate the relationship between H.pylori infection and its distribution of cardia cancer and precancerous lesions. Finally, treated the early cardia cancer and precancerous lesions by EMR and APC, regularly follow up the post-treatment patients, and estimated therapeutic effect.Part 4 A study on the association between IL-8 polymorphisms and susceptibilities to cardia and esophageal squamous cell carcinomaAll the patients, including 320 cases esophageal squamous carcinoma, 340 cases gastric cardiac adenocarcinoma and 404 cases healthy controls ,came from high incidence region of Hebei province. The total patients were diagnosed by histopathology, and detected by 14C-urea breath test. Whereafter, we collected the disease history, personal history, family history,and so on. Genomic DNA was extracted by tiangen kit. IL-8 was genotyped by polymerase chain reaction–restriction fragment length polymorphism (PCR-RFLP). Random sampling of DNA sequencing analysis was used to confirm the results of IL-8 genotyping. Hardy-Weinberg analysis was performed to compare the observed and expected genotype frequencies using Chi-square test. Comparison of the IL-8 genotype and allelotype distribution in the study groups was performed by means of two-sided contingency tables using Chi-square test. The odds ratio (OR) and 95% confidence interval (CI) were calculated using an unconditional logistic regression model and adjusted by age and gender accordingly.Part 5 Application of serum protein fingerprint model in diagnosis of esophageal cancer and gastric cardiac adenocarcinoma and their precancerous lesions in high incidence areaThe patients came from natural population in high incidence area of EC--Ci county and She county from Sep 2004 to Dec 2005. The total patients consisted of 26 cases Esophageal mild dysplasia (EDYSⅠ)、26 cases Esophageal moderate dysplasia (EDYSⅡ)、11 cases Esophageal severe dysplasia (EDYSⅢ)、15 cases Early esophageal cancer (EEC)、36 cases advanced esophageal carcinoma (AEC)、21 cases Gastric-cardia mild dysplasia (GDYSⅠ)、10 cases Gastric-cardia moderate or severe dysplasia (GDYSⅡ~Ⅲ)、34 cases Gastric cardiac adenocarcinoma (GCA) and 38 cases of NOR. All cases were checked by endoscopy with iodine stainning,and confirmed by superior pathologists. Esophageal precancerous lesion and early cancer were all further consultation to confirm the diagnosis. SELDI-TOF-MS and CM10 ProteinChip were used to detect the serum proteomic patterns of the total patients. The Data was analyzed by Zhejiang University Cancer Institute ProteinChip Data Analysis System (ZUCI-PDAS, www.zlzx.net). Each peak in experiment data was estimated by the P value of Wilcoxson T-test. The top ten peaks with the smallest P value were selected for further analysis. Combinations with the highest accuracy in distinguishing different groups of data were selected as potential biomarkers. The SVM model with the highest Youden’s index was selected as the model for detecting esophageal carcinoma and precancerous lesions. The diagnostic model was evaluated and validated by leave one cross validation.ResultsPart 1 Endoscopic screening and analysis of risk factors in high incidence area of esophageal carcinoma1. 3660 cases residents accepted the endoscopic screening. The screening rate was seventy one percent. From above people, we detected 11.9 percent mild dysplasia, 2.1 percent moderate dysplasia, 0.4 percent severe dysplasia, 0.8 percent early carcinoma (carcinoma in situ and intramucosal ), 0.3 percent invasive cancer respectively. In contrast, detection rates of precancerous lesions and cancer from cardia were 1.1, 0.4, 0.2, 0.2 and 0.2 percent respectively.2.With age increasing, the detection rates of precancerous lesions and caner in esophageal and cardia group became gradually higher respectively. There was an association of ascensus trend between the detection rates and age.(esophageal lesions group:χ~2 =130.9,P=0.00;cardia lesions group:χ~2 =40.0, P=0.00)3. One hundred percent early esophageal cancer, 93.8 percent severe dysplasia, 19.9 percent normal and inflammation were positive with iodine staining. There was remarkable difference of positive rate between the ahead and the later two groups.(χ~2=169.8,P<0.01)4. 62 cardiac precancerous lesion are detected by endoscope. In them, 64.9 percent lesions distributed in 12 to 3 clock bits of cardia, and 30.6 percent lesions distributed in other clock bits. There was remarkable difference between the above two groups. Same as above, there was significant difference of clock bits distribution in early cancer groups (P<0.01).5. The detection rates of precursors and carcinomas from the esophagus were higher in male, positive family history of upper gastrointestinal cancer, and smoking group than the opposite groups(χ~2=17.0、22.7、15.8、γ=5, P=0.005、0.000、0.007). There is no correlation between the detection rates and drinking group(χ~2 =4.5,P=0.5).6.The detection rates of precursors and carcinomas from cardia were higher in male, positive family history than the opposite groups(χ~2=34.6、14.3, P=0.000和0.0014). There is no correlation between the detection rates and smoking or drinking(χ~2=8.2、8.0,P=0.16、0.19).Part 2 The intermittent time of Endoscopic screening in high incidence area and the natural history of precancerous lesions from esophagus and cardia1. Periodic screening with shorter interval was performed among 301 cases residents in high incidence area. In above all, 7 cases of SD, 6 cases of cancer in situ and intramucosal carcinoma, 3 cases of invasive cancer were detected by using electronic gastroscope. The longest interval time was 50 months.2. Sojoum time for severe dysplasia(SD) was 13 months after a baseline diagnosis of normal epithelium in 1 subject,7 months after a baseline diagnosis of base cell hyperplasia(BCH) in 1 subject,3 months, 4 months, 4 months, and 10.5 months after baseline diagnoses of mind dysplasia(mD) in 4 subjects,and 12.5 months after baseline diagnosis of moderate dysplasia(MD) in 1 subject.3. Sojoum time for cancer in situ or intramucosal carcinoma was 48 months after a baseline diagnosis of mD in 1 subject,4 months and 13 months after baseline diagnoses of MD in 2 subjects, and 3.5 months, 9 months, and 17.5 months after baseline diagnoses of SD in 3 subjects.4. Sojoum time for invasive cancer was 50 months after a baseline diagnosis of MD in 1 subject, 14 months and 19 months after baseline diagnoses of SD in 2 subjects.Part 3 Epidemic strength, early diagnose and treatment of cardia cancer in high incidence region of esophageal cancer1 Epidemiological investigation of GCA1.1 Incidence and mortality rates of GCA in Shexian countyThe incidence rates of cardia cancer for man and woman in Shexian county from 2000 through 2004 were 69.9 and 41.5, and the mortality rates were 54.3 and 33.2 per 100,000 respectively. Esophageal, cardia and distant stomach cancers put together make up 79.8, 78.7 percent for man, and 68.6, 71.9 percent for woman of the incidence rate or mortality figure for cancer of the whole body.1.2 Compared the detection rates of upper gastrointestinal cancer and precancerous lesions in Shexian , Linxian and Cixian countyEndoscopic survey with iodine staining and biopsy for people aged 40-69 years old could detect 0.6, 0.7 and 0.1 percent intramucosal carcinoma of cardia in Shexian, Linxian and Cixian county respectively. The aboved rates resulted in a ratio of 1:4 to the detection rate of cancer in situ from esophagus. Similar to cancer in situ, detection rate of severe dysplasia from the cardia was also much lower than that from the esophagus, the ratio was 1:4 in Shexian and Cixian county. The detection rates of moderate and mild dysplasia from cardia were similarly much lower than from esophagus.2 H.pylori infection rates and observation about microstructures and ultrastructures of mucosa from cardia lesions2.1 The microstructure of cardiac mucosa with magnifying endoscopeAfter analyzed the relationship between four kinds of mucosal pit patterns and superficial, active and atrophic inflammations, the results indicated that there was significant correlation between these pit patterns and the degree of chronic inflammation. The type of B,C and D displayed in the mucosa of intestinal metaplasia, and type C (66.13 percent) was significantly higher than type B and D in this group. Type C, D and E in 23 cases dysplasia were 48.0, 21.7 and 30.4 percent respectively. The frequencies of type E (64.28 percent) in severe dysplasia group was significantly higher than that in mild and dysplasia dysplasia (P<0.05). All pit patterns of early cardiac cancers were type of E. Analyzed the correlation of dysplasia groups and the pit pattern of muscoal by using rows and column related test (χ~2=10.3, pearson =0.644,P <0.05).2.2The ultrastructure of mucosa with scanning electron microscopeFrom type A to E, the ultrastructure of mucosa observed in low power scanning electron microscope was that the pit was extend and incompleted gradually, from regular punctiform, rod-shape, and en plaque to irregular or disappeard. In high power, we observed that type A was regular circular, made up of two to four cells, all cells which had microvillus were regular and lined up in order. Type B was short rod-shape, there were more cells in the pit, the shape of cells regular and changed little. Type C was arborization, made up of cells more than type B, the pit bent and appeard long slit-shaped ,cells were irregular. Type D was en plaque, the cells were chaotic and degeneration little. Type E was disorganization, cells were degenerate obviously, the microvillus of it were disappeared. Cellula columnoepithelialis of the cryptae like brick wall in sectional view, cells desquamated easily.2.3 Infection rates of H.pylori in cardia lesionsThere was a close relationship between infection rate of H.pylori and cardia cancer and precancerous lesions (χ~2=32.35, P<0.05). The infection rate of H.pylori was the highest in active cardia inflammation. Infection rates of H.pylori in 12 and 6 clock bit were 96.48 and 65.10 percent respectively. There was significant difference in above distribution. (χ~2=108.18, P <0.05).3 Results of the endoscopic therapy in early cardia cancer and precancerous lesions Forty three (78.18 percent) lesions located root of the mucosal fold in cardia (12 to 3 clock bit), twelve (21.82 percent) lesions located other sites. There was statistical significance in above groups. Eleven (73.3 percent) lesions were cut completely with EMR, other four cases remnant lesions were cauterized with APC. The achievement ration of APC were 87.5 and 100 percent respectively. Comparing the pathology of pretherapy and post-treatment, one case intramucosal carcinoma was confirmed postop among four cases preoperative moderate dysplasia, and two cases intramucosal carcinoma were confirmed postop among seven cases preoperative severe dysplasia, and nine cases were bleed, no perforation in all the cases. Regular follow up were performed according to 1,3,6,12 months after treatment.Part 4 A study on the association between IL-8 polymorphisms and susceptibilities to cardia and esophageal squamous cell carcinoma1 Results of PCR-RFLP: The allele gene frequency of IL-8 was 49.6 percent for A in GCA , and 43.1 percent in controls respectively. The differences between the two groups were significantly (χ~2=6.260,P=0.012). However, there were not significantly differences among ESCC and controls (χ~2=6.299 ,P=0.043).2 The genotype distribution of IL-8 in cases and controls: The mutant-type homozygote AA genotype frequency of IL-8 was significant different between GCA and controls [ age, gender, smoking status, family history of UGIC and infection of H.pylori adjusted odds ratio (OR)= 2.014, 95 percent confidence interval(CI)=1.017-3.990]. In ESCC, AA genotype frequency was lower than controls, (15.6percent VS 18.3percent), nevertheless, the differency was not significant(OR=1.901, 95percent, CI=0.952-3.794).3 The stratification analysis of IL-8 Single nucleotide polymorphism and risk factors of cancer3.1 In the smoking status, the AA genotype frequency of positive smoking in ESCC and GCA patients all were not significantly differences than that in controls , and also in the negative groups(P>0.05).3.2 The AA genotype frequency of positive family history of upper gastrointestinal cancer (UGIC) in ESCC (23.1 percent) and GCA (19.8 percent) patients were significantly higher than that in healthy controls (13.8 percent) (P<0.00). Compared with the wild-type homozygote TT, individuals with IL-8-251AA, especially among positive family history of upper gastrointestinal cancer (UGIC), have a higher probability of of ESCC and GCA[the age, gender, smoking stutas and infection of H.pylori ( in GCA groups) adjusted odds ratio (OR)= 2.378 and 14.895, 95percent confidence interval(CI)=1.075-5.258 and 2.889-76.516], but in the negative groups, all the genotype frequencies were not significantly different from that in healthy controls.3.3 Genotype frequency of AA in the group of H.pylori infection positive was significantly higher than that in negative group compared with TT genotype (P=0.017). After adjusted with age, gender, smoking and family history, OR value and 95%CI were 3.520 and 1.249-9.918, respectively. There were no statistics differents in the group of H.pylori infection negative groups.Part 5 Application of serum protein fingerprint model in diagnosis of esophageal cancer and gastric cardiac adenocarcinoma and their precancerous lesions in high incidence area1 Establishment of serum protein fingerprint models in esophageal lesions1.1 We employed SELDI-TOF-MS and support vector machine(SVM)classifier to screen and build nine diagnostic models of EC and precancerous lesions in high incidence area. These diagnostic models are EDYSⅠvs NOR,EDYSⅡvs NOR,EDYSⅢvs NOR,EEC vs NOR,AEC vs NOR,EDYSⅠvs AEC,EDYSⅡvs AEC,EDYSⅢvs AEC and EEC vs AEC.1.2 Among the total, the diagnostic models of EDYSⅠvs NOR,EDYSⅡvs NOR,EDYSⅢvs NOR,EEC vs NOR and AEC vs NOR have specificity (Spe) of 92.11,84.21,81.58,81.58 and 89.47 percent, and have sensitivity (Sen) of 42.31,73.08,45.45,80.00 and 83.33 percent respectively.1.3 The diagnostic models of EDYSⅠvs AEC,EDYSⅡvs AEC,EDYSⅢvs AEC and EEC vs AEC have respectively Spe of 92.31,80.77, 90.91 and 73.33 percent, and have Sen of 80.56,83.33,94.44 and 91.67 percent.1.4 In above-mentioned diagnostic models, the protein peaks (M/Z 4291,4975,5644,5664,and 8775Da) appeared again.2 Establishment of serum protein fingerprint models in cardiac lesions2.1 We employed SELDI-TOF-MS and SVM classifier to screen and build six diagnostic models of GCA and precancerous lesions in high incidence area. These diagnostic models are GDYSⅠvs NOR,GDYSⅡ~Ⅲvs NOR,GCA vs NOR,GDYSⅠvs GCA,GDYSⅡ~Ⅲvs GCA and GDYSⅠvs GDYSⅡ~Ⅲ.2.2 Among the total, the diagnostic models of GDYSⅠvs NOR,GDYSⅡ~Ⅲvs NOR,GCA vs NOR have respectively Spe of 86.84,100 and 94.74 percent, and have Sen of 61.90,90.00 and 88.24 percent.2.3 The diagnostic models of GDYSⅠvs GCA,GDYSⅡ~Ⅲvs GCA have respectively Spe of 90.48 and 80.00 percent, and have Sen of 88.24 and 91.18 percent.2.4 The diagnostic model could differentiate GDYSⅡ~Ⅲfrom GDYSⅠwith a Spe of 100 percent and a Sen of 80.00 percent.2.5 In above mentioned diagnostic models, the protein peaks (M/Z 3271,6891 and 8571Da) appeared again. ConlusionsPart 1 Endoscopic screening and analysis of risk factors in high incidence area of esophageal carcinoma1 The results demonstrated that the screening with endoscopic staining examination could improve early detection of esophageal carcinoma and precancerous lesions.2 There was significant difference about detection rates of early cardia cancer and precancerous lesions between in 12 to 3 clock bits than in other bits. It hinted that we could raise the detection of precances lesion by observing and biopsy in 12-3 sites.3 The results indicated that esophageal and cardia carcinoma developing had gender difference. The people with positive family history of upper gastrointestinal cancer had higher incidence rates than the others with negative history group . There was no remarkable correlation between incidence rate and smoking and drinking.4 Enhancing the compliance and cognition about edoscopic screening in high incidence region, made the high risk people into screening range, could be helpful to improve the detection rates of early carcinoma and precancerous lesions.Part 2 The intermittent time of Endoscopic screening in high incidence area and the natural history of precancerous lesions from esophagus and cardia1 One reasons for interal cases was that significant individual variation existed in the sojourn time, some of them develop rapidly; and the other reason is multifocal carcinogenesis; and the last reason is the false negative results in first time of endoscopic biopsy sampling.2 Periodic screening with shorter intervals should be considered to control the number of interval cases due to rapid development , multifocal carcinogenesis, and false negative results in endoscopic biopsy sampling, change the way of wide-intervaled screenings and then fall behind .3 Scheme of screening in this study was that two or three years to the cases with Normal ,BCH and mD, half or one year to the MD and above. However,this scheme should be further evaluated.Part 3 Epidemic strength, early diagnose and treatment of cardia cancer in high incidence region of esophageal cancer1 The south Taihang mountain region is a high risk area not only for esophagus cancer, but also for cardia cancer. To control upper gastrointestinal tract cancer as a whole in the place, special attention should be paid to the control of cardia cancer by endoscopic screening. The detection rates of early cardia cancer and precancerosis were low, so it is very important to improve the discrimination of above lesions.2 There was a close relationship between type of mucous pits and histopathology by the magnifying endoscope in cardia. The type of pits are useful for biopsy. Type C is a special phenomenon of intestinal metaplasia. The degree of dysplasia gradually aggravated from type C to type E. Type E indicated severe dysplasia and early cancer by magnifying endoscope.3 There was a close relationship between infection of H.pylori and cardia cancer and precancerosis. The infection rate of H.pylori was topmost in active cardia inflammation. It indicated that infection of H.pylori was a risk factor of it, infection rate of H.pylori were higher in 12 clock bit than in 6 clock bit.4 Scanning electron microscope can distinct the fine structure from normal to cardia cancer, and have practical value for the study of mechanism of pit patterns observed by magnifying endoscope.5 EMR and APC are safe and effective methods for treating early cardia cancer and precancerous lesions. The therapeutic patients range should include moderate dysplasia cases. The method of injecting drug in submucosa is one kind of useful method for therapeutic endoscopy.Part 4 A study on the association between IL-8 polymorphisms and susceptibilities to cardia and esophageal squamous cell carcinoma1 People who carry IL-8-251AA genotype have high risk of developing GCA, however, no evidence was found to support that the polymorphisms of IL-8 had relationship with ESCC.2 Family history of upper gastrointestinal cancer (UGIC) significantly enhanced the risk of developing ESCC and GCA, the AA genotype of positive family history of upper gastrointestinal cancer (UGIC) was significantly high risk of developing GCA, the polymorphism of IL-8 alleles may be not associated with the susceptibility of ESCC, but the AA genotype enhanced the susceptibility of ESCC among individuals with UGIC history.3 H.pylori infection positive was risk factor of GCA, the AA genotype of H.pylori infection positive was significantly high risk of developing GCA.4 Smoking was a risk factor of ESCC, but there was no relationship between smoking and GCA, no evidence was found to support that the polymorphisms of IL-8 had relationship with smoking status. Part 5 Application of serum protein fingerprint model in diagnosis of esophageal cancer and gastric cardiac adenocarcinoma and their precancerous lesions in high incidence area.1 In this study, we employed SELDI-TOF-MS to screen and build fifteen diagnostic models of EC, GCA, and their precancerous lesions. In above all, ten diagnostic models could differentiate the above lesions and NOR with high Spe and Sen.2 After further analysed the experimental data, we found that the five protein peaks (M/Z 4291,4975,5644,5664 and 8775Da) had alike function to classify EC and esophageal precancerous lesions from screening crowd, and the three protein peaks (M/Z 3271, 6891 and 8571Da) had similar classification to cardia lesions. The above peaks may be the important biomarkers, and play the significant roles in carcinomatous change.3 Compared with decision tree and artificial neural network classifier, the SVM classifier is a kind of algorithm with many merits. It could guarantee the accuracy and validity of experimental data.4 The above diagnostic models could differentiate EC, GCA, precancerous lesions and NOR well. Their reproducibility and validity should be authenticated through enlarging samples. The valuable protein peaks should be further separation, depuration and identification. It will be helpful to screening in high incidence area.
【Key words】 esophageal carcinoma; gastric cardia cancer; precancerous lesions; high incidence area; endoscopy; screening; early diagnosis and early treatment; SNP; SELDI-TOF-MS; natural history;