节点文献
肺癌诊断的循证医学初步研究
Primary Evidence-Based Study on the Diagnosis of Lung Cancer
【作者】 顾萍;
【导师】 俞守义;
【作者基本信息】 南方医科大学 , 流行病学与卫生统计学, 2007, 博士
【摘要】 20世纪90年代以来,随着循证医学尤其是循证肿瘤学的逐步发展成熟,给肺癌的临床研究带来了快速发展契机。循证医学(evidence-based medicine,EBM)又称求证医学、实证医学。1996年,Sackett将其定义为“有意识地、明确地、审慎地利用当前的最佳证据制定关于个体病人的诊治方案”。与传统的临床医学不同,循证医学是一种理性的医学而不是一种经验的医学,它是建立在现代临床流行病学基础上的现代临床医学的基础课,通过完成对科学性证据的寻求并将其应用于指导临床实践,最终达到使患者得到更好的医疗服务的目的。由于大量设计严谨的临床随机对照研究结果的出现和遵循循证医学基本原则建立的临床指引,近年来有关肺癌的诊治策略研究已将重点从临床症状性肺癌转向无症状、未转移的早期肺癌筛查。目前,肺癌主要的诊断方法有细胞学(痰脱落细胞、支气管肺泡灌洗液、胸水涂片)、影像学(X线、CT、PET)以及组织学(纤维支气管镜活检、肺针吸活检、胸膜活检、纵隔镜活检、剖胸活检、经皮肺活检或手术切除标本活检)检查等。对肺癌诊断具有临床参考价值的检查包括:X线、放射性核素发射计算机辅助断层显像(ECT)、计算机断层扫描(CT)、正电子发射断层显像(PET)等各种影像学技术以及血清肿瘤标记物等。由于每种检查项目都有各自的优缺点,从临床应用实际出发,围绕着肺癌诊断研究,人们将肺癌诊断项目分成一线检查项目和二线检查项目。其中一线检查项目主要由胸部正侧位片、痰细胞学和B超构成,因为它既适用于诊断初诊病人又适用于筛查高危人群,是肺癌诊断中最基本的检查,经济实用,无创伤性,可以反复多次使用。另外,在有条件的情况下,胸部低剂量螺旋CT和纤维支气管镜检查也列入一线检查项目。二线检查项目是在一线检查的基础上,结合各个病人的不同情况,选择某些适当的检查项目,其目的在于进一步明确病变性质或弄清有无远处转移扩散,以确定治疗方案,具体包括胸部或脑部磁共振成像(MRI)、PET、胸腔穿刺术、颈部淋巴结穿刺或活检术、经皮肺穿刺术、纵隔镜检查术、胸腔镜检查术、全身骨扫描、血清肿瘤标记物等。本文在系统文献调研和专家咨询的基础上,通过确立临床常用肺癌诊断试验与患者初诊时的主诉指标,设计了“肺癌病例诊断指标采集表”,并利用数据管理软件Epidata3.1建立了“肺癌病例诊断指标采集表”录入平台,对所收集的1100个原发性肺癌病例以及作为对照的1000个肺良性疾病病例(肺结核、肺炎、肺炎性假瘤)进行了相关数据录入,完成了“肺癌病例诊断数据库”的构建工作。在此基础上,利用SPSS13.0软件,首先参照《中国肺癌杂志》2003年第5期上发表的《肺癌的诊断和分期临床指引》,对肺癌常用诊断试验的临床应用情况进行了初步评价;其次,根据循证医学在诊断试验中的有关应用原理,对胸部X线、CT和纤维支气管镜三种方法在肺癌诊断中的准确性以及联合应用情况下的优势互补性进行了探讨;最后,针对肺癌诊断的主观依据(患者初诊时的主诉内容,包括一般项目、家族史、既往史、吸烟史、临床症状与体征等)与客观依据(医生采用的诊断试验),对肺癌临床诊断的主要影响因素进行了单因素Mantel-Haenszel以及多因素非条件logistic模型分析,初步构建了肺癌诊断模型,同时以Microsoft VB6.0软件和C语言为平台开发了开发了肺癌个性化诊断系统,采用SQL Server软件建立诊断病例数据库,初步达到了根据患者个体特征采用不同的诊断方案预测患者患病的概率,并自动筛选和优化诊断方案的目的。具体结论如下:1.肺癌患者临床诊断试验的循证分析采用配对McNemarx~2检验,参照《肺癌的诊断和分期临床指引》,对611例肺癌常用诊断试验的临床应用情况进行的初步评价结果发现:1.1肺癌患者最常见的临床症状是咳嗽(400/611),持续两周以上的症状包括咳嗽93.8%(375/400)、血痰86.1%(130/151)、胸痛85.5%(142/166)和气促87.1%(88/101)等。1.2病理类型男性以鳞癌(218/462)、女性以腺癌(102/149)多见;约9.5%(58/611)的肺癌患者诊断时没有任何症状。1.3痰细胞学检查阳性检出率为41.3%(100/242);血清CEA阳性检出率为48.9%(107/219);胸部X线+胸部CT+血清CEA和胸部X线+胸部CT+痰细胞学组合做平行试验的阳性检出率分别可达98.8%(81/82)和97.8%(88/90),但由于本组病例同时采用胸部X线+胸部CT+血清CEA检查组合的有13.4%(82/611),采用胸部X线+胸部CT+痰细胞学试验检查组合的有14.7%(90/611),这可能说明临床诊断实践过程中,存在着诸如病程长短、医生经验以及过分依赖高新技术等因素的影响,有待今后进一步开展这方面的研究。1.4共有81.8%(500/611)的肺癌患者进行了胸部X线检查,有74.6%(456/611)患者行正侧位片检查,使得阳性检出率从原来的71.4%(357/500)提高到73.5%(335/456);CT的检查使用率虽然只有45.8%(280/611),但阳性检出率却达到了97.9%(274/280)。另外,在335例行胸部正侧位X线检查疑似病例中,虽然只有36.7%(123/335)的患者做了CT检查,但阳性检出率却高达97.6%(120/123)。提示为了弥补X线检查分辨率低的不足,提高定位的准确度,有助临床TNM分期,临床经胸部正侧位片疑诊肺癌的病例,要常规进行胸部CT检查。1.5有71例非小细胞肺癌(Non Small Cell Lung Cancer,NSCLC)患者进行了PET检查,其中ⅢA~Ⅳ期患者共50例,占70.4%(50/71),与《指引》中PET的临床应用建议基本相符,说明PET这种临床昂贵检查资源在肺癌诊断实践中得到了较合理的使用。1.6纤维支气管镜的使用率为44.2%(270/611),阳性率为93%(251/270),如若与胸部X线及CT开展平行试验,则检查的阳性率可达100%。另外,本组病例中,Ⅰ~ⅢA期患者353例,其中有89例做了纤维支气管镜检查,检查使用率为25.2%(89/353),这与《指引》中提出的“临床怀疑Ⅰ~ⅢA期的肺癌病例,应常规进行纤维支气管镜检查”的建议不符。提示今后肺癌临床诊断过程中,针对疑似Ⅰ~ⅢA期的肺癌病例,在考虑患者依从性的前提下,注意纤维支气管镜的检查和使用力度,努力降低检查的漏/误诊率。1.7本组病例虽然采用CT定位经皮肺穿刺活检技术的病例只有15.4%(94/611),但阳性检出率却达到了96.8%(91/94)。经支气管肺活检检查的使用率为20.6%(126/611),阳性检出率为88.9%(112/126)。通过对诊断试验检查结果的一致性评价,发现该项检查与CT定位经皮肺穿刺活检对肺癌患者的阳性检出率具有一致性,这说明临床实践中存在着过度诊断问题。本组病例中,共有9例患者同时实施了上述二项诊断试验,可能在一定程度上给患者带来不必要的痛苦和经济负担。2.胸部X线、CT和纤支镜对肺癌诊断的应用研究根据循证医学在诊断试验中的有关应用原理,通过回顾性分析140例原发性肺癌和95例肺良性疾病患者的临床资料,对胸部X线、CT和纤维支气管镜三种方法在肺癌诊断中的准确性以及联合应用情况下的优势互补性进行了探讨,结果发现:2.1 X线检查对中央型肺癌的符合率为62.5%(45/72)、周围型肺癌的符合率为78%(32/41)、细支气管肺泡癌的符合率为66.7%(2/3),也就是说超过60%的肺癌患者仅依据检查简单、价格低廉的X线检查即可诊断。此外,尽管行X线检查前临床估计某疑似病例的验前概率只有30%,但当检查结果(+),该病例的患病可能性将达71.67%,说明X线的确是临床诊断肺癌的首选方法。2.2由于CT检查的灵敏度和阳性预测值分别达到了92.1%(129/140)和90.8%(129/142),因此上面提到的某病例,即使行X线检查前其验前概率为30%,当X线检查结果(+),其验后概率为71.67%,若与CT作序列试验,则验后概率将提高到94.44%,大大提高了肺癌诊断的准确度。另外,CT检查对周围型肺癌诊断的符合率95.1%(39/41)高于中央型肺癌的88.9%(64/72),也就是说对中央型肺癌的漏诊率较高。2.3纤维支气管镜检查对中央型肺癌的诊断符合率高达95.8%(69/72)。另外,在三项诊断试验中,纤维支气管镜的灵敏度、特异度、阳性预测值和阳性似然比最大,说明相对于X线和CT,纤维支气管镜检查对肺癌诊断具有较高的临床应用价值。纤维支气管镜若与X线作序列试验,阳性似然比将达到173.5,则在验前概率为30%的前提下,验后概率将从原来的71.67%提高到98.67%。这对于临床肺癌诊断实践中,坚持循证医学从简单到复杂、从无创到有创的疾病诊断原则有着重要的指导意义。3.肺癌诊断模型与个性化诊断系统初步研究通过设计以医院人群为基础的病例对照研究,以997例原发性肺癌和886例肺良性疾病患者为研究对象,针对肺癌诊断的主观依据与客观依据,对肺癌临床诊断的主要影响因素进行的单因素Mantel-Haenszel分析结果发现:血痰、胸痛、气促、消瘦、乏力、食欲减退、肿瘤局部扩散和远处转移的OR值较大,且P值小于0.05,病例组与对照组有显著性差异。另外,由于吸烟、环境污染、职业因素、营养和饮食、慢性肺部疾病、肿瘤家族史和遗传易感性是主要的肺癌危险因素,并且对肺癌诊断具有临床参考价值的检查主要包括各种影像学技术以及血清肿瘤标记物等,因此,具有统计学意义的指标,如住地、吸烟史、痰细胞学、X线、CT、血清CEA、纤维支气管镜、纤维支气管镜直接征象、经支气管肺活检和CT定位经皮肺活检等可作为肺癌诊断模型的有效指标,在此基础上,采用非条件logistic回归模型初步建立了肺癌诊断的模型,并以MicrosoftVB6.0软件和C语言为平台开发了肺癌个性化诊断系统,初步实现了在不同验前概率条件下,根据患者个体特征采用不同的诊断方案预测患者患病的概率,并进一步筛选和优化诊断方案的目的。通过对113例外对照病例的再验证观察,发现所构建的诊断模型对肺癌患者的诊断符合率为98.8%(81/82),所开发的肺癌个性化诊断系统对肺癌患者的诊断符合率为81.7%(67/82)。这对于提高肺癌临床诊断操作的规范化水平,降低肺癌的漏/误诊率将具有积极的意义。
【Abstract】 Since 1990s, the clinical research of lung cancer has obtained a rapid developingchance with the development of evidence-based medicine (EBM), especiallyevidence-based oncology. In 1996, Sackett named EBM as: "the conscientious,explicit, and judicious use of the current best evidence in making decisions about thecare of individual patients". Different from the traditional clinical medicine, EBM is akind of rational medicine non experimental medicine. It is established on the basis ofmodern clinical epidemiology. EBM can make patients get better care by seeking andusing scientific evidence in clinical practice.Recently, the emphasis on the strategy about diagnosis and treatment of lungcancer has been transmitted to screening for early lung cancer according to clinicalsymptoms of lung cancer since the appearance of a large amount of research results ofRCT and clinical guidelines based on the primary principles of EBM.At present, the major diagnostic methods of lung cancer refer to Cytology, Imageand Histology, etc. The diagnostic tests which are beneficial to the diagnosis of lungcancer include X-ray, ECT, CT, PET and serum tumor markers, etc. Considering eachkind of tests has its prons and cons, people have divided all the diagnostic tests intofront-line and second-line tests according to the need of clinical practice. Thefront-line tests include chest PA & LAT, sputum cytology and B-ultrasound. They are the most basic diagnostic tests of lung cancer because they are adaptable to both firstvisit patients and screening for the high risk crowd. In addition, low-dose spiral CTand fibrobronchoscope can be added into front-line tests under some condition.Second-line tests refer to selecting some proper tests according to different clinicalcondition of each patient based on the results of front-line tests. Their action aims atmaking clear lesion’s characteristics or whether or not the occurrence of distantdiffusion, and deciding the treatment strategy. Second-line tests mainly include chestor brain MRI, PET, Thoracocentesis, needle biopsy in cervical lymph node,mediastinoscopy, SPECT and serum tumor markers, etc.The author had designed "diagnostic items collecting table of lung cancer cases"on the basis of systematic literature review and experts consulting by surveyingroutine clinical diagnostic tests of lung cancer and chief complaints of first visitpatients, and set up "data input system" by using data management software ofEpidata 3.1. Furthermore, Diagnostic Database of of Lung Cancer Cases had beenestablished by inputting the related information of 1100 primary lung cancer casesand 1000 benign pulmonary lesion cases as controlled. In the end, diagnostic instanceof lung cancer cases had been retrospectively analyzed and studied from followingthree aspects by using SPSS 13.00 software:1. Referring to "Clinical Guideline of Lung Cancer Diagnosis and Staging"published on "Chinese Journal of Lung Cancer" issue 5, 2003, we had evaluatedclinical application of routine diagnostic tests of 611 lung cancer cases by usingMcNemarx~2 test. The results showed as follows:1.1 Cough was the most routine clinical symptom of lung cancer patients. Atthe same time, symptoms of cough, blood sputum, chest pain and shout breath lastedabove two weeks in terms of 93.8%(375/400), 86.1%(130/151), 85.5%(142/166) and 87.1%(88/101) lung cancer patients respectively.1.2 There were more male patients in squamous carcinoma while more femalepatients in adenocarcinoma; pathological types of middle-age and elderly patientsabove 40 were mainly composed by squamous carcinoma and adenocarcinoma; about9.5% lung cancer patients had no symptoms during diagnostic period.1.3 Positive rate of sputum cytology is 41.3%, serum CEA 48.9%; detectingrate of X-ray plus CT plus CEA group is only 13.4% while X-ray plus CT plussputum cytology group 14.7%. Therefore, advanced survey would be done todemonstrate whether factors such as the course of disease, doctor’s experience andover depending on high techniques influence the selection of diagnostic tests or not.1.4 81.8%(500/611) of lung cancer patients had carried through X-rayinspection, among of them, 456 cases took chest PA & LAT, thus, positive rate hadbeen improved to 73.5% from 71.4%; although CT inspecting rate was 45.8%,positive rate of its reached to 97.9%. In addition, only 36.7%(123/335) X-raysuspected cases had carried through CT, but positive rate got to 97.6%(120/123),hinting that CT inspection should be selected in clinical practice in terms of X-raysuspected cases.1.5 Since PET inspection had been selected by 71 NSCLC patients among ofthem 50 beingⅢA-Ⅳ, in accordance with the suggestion of the guideline mentionedabove, it could be concluded that PET had been applied rationally in routine clinicalpractice.1.6 Inspecting rate of fibrobronchoscopy was 44.2%, positive rate 93%, ifdeveloping parallel test with X-ray and CT, then, positive rate 100%. Additionally,only 89Ⅰ-ⅢA patients had carried through fibrobronchoscopy, and it was differentfrom the suggestion mentioned above, therefore, it was demonstrated thatfibrobronchoscope should be used byⅠ-ⅢA patients in order to reduce omission diagnostic rate and mistake diagnostic rate.1.7 Although only 15.4% cases had carried through CT locating percutaneousneedle biopsy, its positive rate got to 96.8%, while inspecting rate of bronchoscopicneedle biopsy was 20.6%, positive rate 88.9%. The evaluation results of consistencyof these two tests showed that both of them had consistency of positive rate whenused in lung cancer patients. Advanced study was still needed in order to demonstratewhether there was overdiagnostic problem in clinical practice or not.2. According to principles of EBM used in diagnostic tests, clinical informationof 140 lung cancer cases and 95 benign pulmonary lesion cases had beenretrospectively analyzed, and the accuracy and strength supplement of X-ray, CT andfibrobronchoscope in the diagnosis of lung cancer had been studied. The resultsshowed as follows:2.1 The accurate rate of X-ray to central lung cancer was 62.5%, peripherallung cancer 78%, and thin bronchuses lung bubble cancer 66.7%, that was, over 60%lung cancer patients could be clinically diagnosed only by X-ray. In addition,although pre-test probability of a suspected case was only 30%, when X-ray resultwas(+), his post-test probability reached to 71.67%, thus, it could be concluded thatX-ray was surely the first selected test in the diagnosis of lung cancer.2.2 Since the sensitivity and positive predictive value of CT reached to 92.1%and 90.8% respectively, even if the pre-test probability of a suspected case was 30%before taking X-ray, when X-ray result was(+), the post-test probability was 71.67%,if developed serial test with CT, the post-test probability reached to 94.44%. Thediagnostic accuracy of lung cancer had been greatly improved. Additionally, theaccurate rate of central lung cancer was higher than that of peripheral lung cancer byCT scans, that is, omission diagnostic rate of CT to central lung cancer was increased.2.3 The accurate rate of central lung cancer was 95.8% by fibrobronchoscopy. Among the three diagnostic tests, the sensitivity, specificity, positive predictive valueand positive likelihood ratio of fibrobronchoscopy were the most highest. It could beconcluded that fibrobronchoscopy had a higher value in the diagnosis of lung cancercompared with X-ray and CT. If fibrobronchoscopy developing serial test with X-ray,its positive likelihood ratio would reach to 173.5, led to post-test probabilityincreased to 98.67% from 71.67% under the condition of pre-test probability being30%. This result would be important to the instructing utility of diagnostic principleof EBM being from simple to complicated, from unwound to wound in clinicalpractice.3. According to the subjective and objective diagnostic evidences of lung cancer,single variable Mantel-Haenszel analysis of major diagnostic influencing factors wasimplemented. The results showed that: blood sputum, chest pain, shout breath, thin,feeling tired, anorexia, local diffusion and distant metastasis had a higher OR valuerespectively, P value was<0.05. There was a significant difference between casegroup and controlled group. In addition, smoking, environment pollution, occupation,nutrition and diet, chronic pulmonary disease, family history of tumor, genetic riskwere the main risk factors of lung cancer, plus the clinical valued diagnostic tests oflung cancer including all the imaging techniques and serum tumor markers, thus,resident, smoking history, sputum cytology, X-ray, CT, serum CEA,fibrobronchoscope and its direct signs, bronchoscopic needle biopsy and CT locatingpercutaneous needle biopsy could be designated as the valued indexes of diagnosticmodel of lung cancer. Furthermore, individual diagnostic system of lung cancer hadbeen developed according to Microsoft VB6.0 and C language in order to predictpost-test probability of patients under different pre-test probability. After theprospective observation of diagnostic model and automatic diagnostic system of lungcancer by using 113 cases, it could be obtained that both of them had a higher accurate rate to lung cancer patients and were significant to the diagnosticstandardization of lung cancer and to the decrease of omission diagnostic rate inclinical practice.All in all, this paper had firstly evaluated clinical application of routinediagnostic tests of 611 lung cancer cases by referring to "Clinical Guideline of LungCancer Diagnosis and Staging". Secondly, the accuracy and strength supplement ofX-ray, CT and fibrobronchoscope in the diagnosis of lung cancer had been studiedaccording to principles of EBM used in diagnostic tests. Thirdly, single variableMantel-Haenszel analysis of major diagnostic influencing factors was implementedaccording to the subjective and objective diagnostic evidences of lung cancer; anddiagnostic model and individual diagnostic system of lung cancer had been designedand developed respectively. In the end, the aim of screening and optimizingdiagnostic strategy had been primarily reached on the basis of individualcharacteristic of patients.
【Key words】 Lung Neoplasm; Evidence-based Diagnostic Analysis; Evaluation of Diagnostic Test; Diagnostic Model; Individual Diagnostic System;