An Exploratory Research Of The EBUS Models On Diagnosis Of Peripheral Lung Cancer

BackgroundWith the rapid development of cancer research, We had made much progess in chemotherapy radiotherapy,immunotherapy and so on.More and more cancer have been effective perished ealier.Howerever,the therapeutic efficacy is still unsatisfactory so far.If we do not take effecetive contral,The population will reach to lmillion at the rate of26.9%each year till the end of2025,as the biggest lung cancer country.Peripheral lung cancer and pneumonia account to63.4%present as the image of nodules or mass,followed by tuberculosis,The three kind lesions are diffcult to identify especially for the spherical nodules of3cm or less, the misdignosis rate can be15%-40%,So people are pursuing more easy and noninvasive method make up for this situation.High-frequency ultrasonography with a miniature probe through a flexible bronchoscope is termed "endobronchial ultrasonography"(EBUS).It is an envolving technology,first introduced in the early1990s,which has enhanced dignostic bronchoscopy by allowing the bronchoscopistsview to extend beyond the airway to peribronchial structures and the peripheral lung nodules;provided physicians a chance to espy the lesions from inside and discover more information other than that obtained through radiology;even tissue15to20mm from the probe could be visualized clearly.However,more and more specialists found.The clinical data,including age,cigarette-smoking status, history of extrathoracic malignant neoplasm and secrum tumor markers concentration is also important in discrimating malignant pulmonary lesions.CT examzation is still the most sensitive,non-invasive and widely used method in detecting PPLs and solitary pulmonary nodes,especially with the dynamic enhancement and high resolution CT reflecting internal blood flow, characteristics,such as spiculation.lobulation,pleural indention and vacuole sign appearing.Our destination is to seek a simpler method in discrimating neoplasm and nonneoplasm peripheral pulmonary lesions based on images of EBUS other than radiology or CT with increasing minimal operating time and low cost.Objective1、To establish models to estimate the probability of maligancy in peripheral pulmonary lesions basedon endobronchial ultrasonographic features、HRCT images and clinical data.2、To compare its accuracy with the model established with the HRCT image characteristics.providing a new effective tool to differentiate malignant lesions from benign.The research objects1、Cases source The subjects in the study were patients who underwent bronchoscopy and EBUS to assess peripheral lung lesions in Respirary medicine.Thoracic surgery,Medical Oncology in Guang Dong General Hospital,Guangzhou,Guangdong province, Between September1st2010and October30th2011.Every patient who comply received informed consent with detailed illustrations before EBUS examinzation.2、The inclusion criteria (1) age18-80years old;(2) The lesions located below segmental bronchus through the X rays or HRCT images;(3) Have signed the informed consent.3、Exclusion criteria (1) hemorrhagic tendency or coagulation disorders;(2) Severely heart or lung condition intolerance to check;(3) No informed consent.Methods 1、EBUS examzation、variables selection and assignmentHeare rate and oxygen saturation were montiored in each patient by a pulse oximeter during the procedure. After video bronchoscopic routine examation,the miniature probe was inserted into bronchi that were suspected to lead to the lesions visualized by plain radiography or CT scanning.Then adjust gain, sensitivity time control, contrast, and image display range parameters, make the image imaging clearly. captured image was saved in a compture storage system,until3to5hardcopies were storaged. If the inferences were identical, the hard copies were sent to another independent reviewer without mention of previous results.the patient was included in the analysis only if the three inferences unanimous.At last,150patients were enrolled.2、Variables of HRCT selection and assignmentTwo independent experienced radiologists analyzed the HRCT images without mention of previous pathology results. The lesion position, boundary, size, vacuole sign, enhanced CT value.calcification, Vascular cluster, spiculation,lobulation,pleural indention bronchial and lymph nodes change were recorded,and150patients were finally enrolled.3、The clinical data variables selection and assignmentThe clinical data,including age,cigarette-smoking status,history of extrathoracic malignant neoplasm and secrum tumor markers concentration were collected,if this condition exists,it notes1,otherwise0.4、The pathologic result record and assignmentThrough bronchiallung biopsy and cytology, phlegm off cytology,percutaneous lung biopsy, thoracoscope biopsy, open thoracic exploration, surgical resection for clear pathology.Record the pathology results of enrolled patients, benign lesions noted for0, malignant lesions noted for1.5、StatisticsComparisons of the clinical characteristics and EBUS findings between different groups were made using the chi square test or Fishers exact test(if appropriate).The prediction model was identified from binary logistic regression analysis. A LR-Forward procedure was used to select independent variables from the statically significant variables in univariate analysis.The performance of the prediction model was evaluated by calculating the prediction accuracy,the sensitivity and specificity was performed on the basis of prediction model and pathology; A p value less than0.05was considered statistically significant.Results(1) Through the univariate analysis,one clinical data.five distinct EBUS image patterns:①the distinct borderline;②internal irregular Anechoic area;③heterogeneity;④air bronchogram (with or without internal hyperechoic dots and linear arcs);⑤concentric circle; five HRCT image patterns:①size;②enhanced CT value;③spiculation;④lobulation;⑤pleural indention were demonstrated significant difference by univariate analysis,the others were not and excluded.(2) According to the result of binary multivariable logistic regression analysis,we combined clinical data and EBUS images to predict the presence of malignancy.The equation of malignancy probability for any patient was:P=1/[l+e-z] Z=(-2.986+1.552borderline+1.616heterogeneity-2.011air bronchogram+1.748anechoic area+1.993smorking+2.293CEA)(e is for constants).(3) According to the result of binary multivariable logistic regression analysis,we combined clinical data and HRCT images to predict the presence of malignancy.The equation of malignancy probability for any patient was:P=1/[l+e-z]. Z=(-5.650+2.799size+1.616spiculation+1.871lobulation+1.520pleural indention+2.013smorking+1.651CEA)(e is for constants)(4) The accuracy of the model only with EBUS images was very good with sensitivity of81.8%, a specificity of70.0%,an accuracy of88.4%;The accuracy of the model obtained with clinical data and EBUS images was very good with a sensitivity of89.6%, a specificity of72.4%,an accuracy of92.6%.(5) The accuracy of the model obtained only with HRCT images was very good with a sensitivity of79.2%, a specificity of72.4%,an accuracy of84.9%.The accuracy of the model obtained with clinical data and HRCT images was very good with sensitivity of84.4%, a specificity of77.6%, an accuracy of90.3%Conclusions(1) Endobronchial ultrasonography is a safe and practical examination method. It can provide physicians a chance to observe the lesions from inside and discover more information than that obtained through radiology in discriminating peripheral pulmonary lesions.(2) Multivariable logistic regression model discriminated between benign and malignant lesions is better than univariable logistic regression. Combined multiple variables predictiing the malignant probability could raise their sensitivity, specificity and accuracy.(3) Combining image results with the clinical data to estamblish binary logistic models to predict the probability of malignancy,can improve the sensitivity.specifity and accuracy,which is accordance with clinical diagnosis mentality.(4) Compared with the traditional HRCT technique, EBUS has higher sensitivity and accuracy, it is worth promotion in clinical.