| Part â… 18F-FDG PET/CT features of primary lesion in prediction of lymph node metastasis and distant metatasis in lung cancerBackgroud Lung cancer is the leading cause of morbidity and mortality in malignant tumors around the world. Local recurrence, lymph node metastasis and distant metastasis are major causes of leading to treatment failure. The obviouse heterogeneity in different pathological types of lung cancer results in obviouse d ifferent biology behavior. The long-term prognosis is not same even if the same TNM staging of lung cancer. The ability of uptaking FDG in primary lung tumo rs is closely relative to some biological indicators, such as tumor cell proliferatio n activity, differentiation degree and microvascular density in tumors, those are k nown to predict lung cancer metastasis potency. The level of uptaking FDG in p rimary lung tumor perhaps can provide objective imaging information and becam e biological marker to evaluate the lung cancer metastasis potency.Purpose To select primary lung tumor features in18F-FDG PET/CT associate d to lymph node metasitasis and distant metastasis of lung cancer by a retrospec tive analysis.To establish mathematical model for predicting lymph node metasitas is and distant metastasis of lung cancer based on primary lung tumor features in18F-FDG PET/CT. To evaluate the diagnostic efficiency of the mathematical mod el by a perspective study.Meterials and Methods One hundred and thirty-one patients (134lesions) w ith lung cancer confirmed by pathological examination underwent18F-FDG whole body PET/CT scans and were reviewed from June2010to October2012. Sixty-four primary lesions without lymph node metasitasis and distant metastasis are d ivided into the no metastasis group. Seventy primary lesions with lymph node m etasitasis or distant metastasis are divided into the metastasis group. Seventy-nine patiens (82lesions) with lung cancer confirmed by pathological examination un derwent18F-FDG whole body PET/CT scans and obtained clear diagnosis about with metatasis or not. They were incorporated in the prospective study from Nov ember2012to October2013. The maximum long diameter length and short dia meter length crossing a tumor section in CT lung window and mediastinal windo w were measured respectively. The maximum long diameter length was regarded as the size of tumor. The lesion density was measured in a ROI covering70%t umor section. TDR (Tumor Disappearance Rate) was calculated by the formula,1-long diameter lengthx short diameter length in CT mediastinal window/long di ameter lengthx short diameter length in CT lung window. In Siemens syngo statio n, byusing True D sofeware to draw3D ROI, the maxium standarded uptake val ue of pixels in ROI was regarded as tumor SUVmax and the sum of all pixel v olume whose SUV larger than or eaqual to2.5was regarded as MTV (Metaboli c Tumor Volume, MTV). Normal distribution measurement data were experessed with mean^standard deviation, Non-normal distribution measurement data were ex peressed with the median and quartiles, M (Q1～Q3). In the retrospective analys is, the differences in patients age, gender, lung cancer lesion location (different1ung lobe), image type (central type and periphery type) and pathological types (a denocarcinoma and squamous cell carcinomas) between the no metastasis group a nd metastasis group were tested by/test or chi-square test. The differences in p rimary tumor size, density, TDR, SUVmax and MTV between the no metastasis group and metastasis group were tested by Mann-Whitney test. The statistical sig nificant variables in predicting lymph node metastasis and distant metastasis were selected and multi-viriables binary logistic regression (forward stepwise) were an alysed to obtain a mathematical model,P<0.05is the incorporation criteria, P>0.1is the exclusion criteria. In the prospective study, the pathology results, other i maging evidence and follow-up results were adopted as the gold standard to dig nose metastasis or not. The diagnostic efficiency of the mathematical model was evaluated on predicting lymph node metastasis and distant metastasis in lung ca ncer.ResultsIn the retrospective analysisThe age in the no metastasis group was63.6±9.53years old and63.2±8.53y ears old in the metastasis group. There was no significant difference between the two group (t=-0.284, P>0.05). The metastasis probability in different lesion loca tion (different lung lobe) and pathological type (adenocarcinoma and squamous c ell carcinomas) did not show statistical differences between the two groups(χ2=2.076,1.581, P>0.05). The metastasis probability in men was66%and40%in women(χ2=7.750,P<0.01). The metastasis probability in central type lung caner was74%and49%in periphery type(χ2=4.081,P<0.05). The tumor size, density, TDR, SUVmax and MTV in the no metastasis group were1.90(1.40-2.40)cm,-1.0(-139.0～28.5)HU,0.59(0.29～1.00),3.20(1.55～9.25) and0.51(0.00～3.47)cm3respectively. In the metastasis group, they were3.10(2.40～4.00)cm,30.0(25.0-38.0)HU,0.27(0.14～0.47),13.70(10.15～19.05) and12.93(4.86～26.17)cm3respectiv ely. There were significant differences between the two groups(Z=-5.148,-4.724,-4.616,-6.270,-6.947,P<0.01). Regarding the patient gender, primary lung tu mor imaging type and the tumor size, density, TDR, SUVmax and MTV as the independent variables, the results of logistics regression analysis showed that pri mary tumor TDR and SUVmax were incorporated into the regression equation(W ALD=7.627,0.067,P=0.006,0.067). OR were0.015and1.043respectively. The mathematical model based on primary lung tumor fetures in18F-FDG PET/CT was stated by P=ey/(1+ey), y=1.468+0.042×primary tumor MTV-4.212×primary t umor TDR; P was expressed by the probability to lymph node metastasis and di stant metastasis; The unit of MTV is cm3.In the prospective analysesSeventy-nine patients (82leisions) were satisfied with the inclusion criteria. J udged by the "gold standard", the number of lesions without metastasis was35and with metastasis was47. Judged by mathematical model, the number of lesio ns without metastasis was38and with metastasis was44. The specificity, sensiti vity, accuracy, positive predictive value and negative predictive value were80%,79%,79%,84%and74%.ConclusionPatient age, primary lung cancer lesion location (deffrent lung lobe) and path ological types (adenocarcinoma and squamous cell carcinomas) cannot predict ly mph node metastasis and distant metastasis in lung cancer. Male patients with lu ng cancer than female patients, the central type lung cancer than peripheral type develop easierly lymph node metastasis and distant metastasis.Primary lung tumor size, density, TDR, SUVmax and MTV are predictive fa ctors to lymph node metastasis and distant metastasis in lung cancer. Larger tu mor size, higher tumor density, smaller TDR and higer SUVmax and MTV mea n the higer probability to lymph node metastasis and distant metastsis in lung c ancer. TDR and MTV were major predictive factors.The mathematical model based on primary lung tumor fetures in18F-FDG P ET/CT to prediction lymph node metastsia and distant metastasis is stated by P=ey/(1+ey), y=1.468+0.042×primary tumor MTV-4.212×primary tumor TDR. The diagnostic efficiency is good. Part â…¡Estabilishment of a Mathematical Modle Based on18F-FDG PET/CT for Lymph Node Metastasis Diagosis in Lung CancerBackgroud Lung cancer is the leading cause of morbidity and mortality in malignant tumors around the world. Lymph node metastasis status in lung cancer is very important to treatment decisions and prognosis evaluation. Diagnosis of lymph node metastasis based on lymph node size in chest CT scan or on the gl ucose metabolism level in PET scan perhaps is false negative or false positive. The level of FDG intake in primary lung lesion is a predictor of lymph node m etastasis. Combination the features of both primary tumors and lymph nodes in18F-FDG PET/CT may improve the accuracy in diagnosis of lung cancer lymph n ode metastasis.Purpose To select features of primary lung tumors and lymph nodes in18F-FDG PET/CT associated to lymph node metasitasis by a retrospectively analysis. To establish mathematical model for diagnosis of lymph node metasitasis in lung cancer based on18F-FDG PET/CT.Meterials and Methods One hundred and fifty-five patients (161lesions) wi th lung cancer confirmed by pathological examination underwent18F-FDG whole body PET/CT scans and were reviewed from June2010to October2013. Ninety-nine primary lesions without lymph node metasitasis are divided into the no ly mph node metastasis group. Sixty-two primary lesions with lymph node metasitas is are divided into the lymph node metastasis group. According to the N stage, t he numbers in N0, N1, N2and N3group are99,14,23and25respectively. T hree hundred and fifty-one lymph nodes were observed and measured in PET/CT scan in the correspondent pathological examination region. The numbers in the non-metaststic lymph node group and metaststic lymph node group are274and77respectively. The maximum long diameter length and short diameter length cr ossing a tumor section in CT lung window and mediastinal window were measur ed respectively. The maximum long diameter length was regarded as the size of tumor. The lesion density was measured in a ROI covering70%tumor section. TDR (Tumor Disappearance Rate) was calculated by the formula,1-long diame ter lengthxshort diameter length in CT mediastinal window/long diameter length×short diameter length in CT lung window. In Siemens syngo station, by using True D sofeware to draw3D ROI, the maxium standarded uptake value of pixel s in ROI was regarded as tumor SUVmax and the sum of all pixel volume who se SUV larger than or eaqual to2.5was regarded as MTV (Metabolic Tumor V olume, MTV). The maximum short diameter length was measured in CT mediast inal window and regarded as lymph node size. In Siemens syngo station, by usi ng True D sofeware to draw2D ROI in every lymph node sections, the maximu m SUV in pixels was regarded as lymph node SUVmax. Lymph nodes with calc ification mottling or fatty, with higher density than soft tissue, with soft tissue d ensity and with necrosis were scored to0,1,2,3, respectively. Lymph nodes wi th clear margin were scored to0, otherwise1. Lymph nodes with a symmetry di stribution in PET was scored0, otherwise1. Normal distribution measurement da ta were experessed with mean+standard deviation, Non-normal distribution measur ement data were experessed with the median and quartiles, M (Q1～Q3). The di fferences in patient age, gender, lung cancer lesion location (different lung lobe), image type (central type and periphery type) and pathological types (adenocarcin oma and squamous cell carcinomas) between the no lymph node metastasis grou p and lymph node metastasis group were tested by t test or chi-square test. The differences in primary tumor size, density, TDR, SUVmax and MTV in the two groups and every other alternate N groups were tested by Mann-Whitney test. The differences in lymph node size and SUVmax were tested by Mann-Whitney test between the non-metastatic lymph node and metastatic lymph node groups. T he differences in lymph node density, margin and distribution score were tested by chi-square test. The statistical significant variables in predicting lymph node metastasis were selected and multi-viriables binary logistic regression (forward ste pwise) were analysed to obtain a mathematical model,P<0.05is the incorporatio n criteria, P>0.1is the exclusion criteria. The receiver-operating characteristic (R OC) curves were obtained by lymph node size, lymph node SUVmax and mathe matical model. The differences of the area under curve (AUV) were tested. Acco rding to maximum Youden index principle, the cutoff values were determined for lymph node size and SUVmax.ResultsFactors relevant to lymph node metastasisThere were no significant differences in patient age, lung cancer lesion locati on (different lung lobe) and pathological type (adenocarcinoma and squamous cel1carcinomas) between the no lymph node metastasis group and lymph node met astasis group (t=-0.310,χ2=2.664,1.360,P>0.05). The metastasis probability in men was51%and26%in women(χ2=10.900,P<0.01). The metastasis probabilit y in central type lung caner was68%and34%in periphery type(χ2=9.475,P<0.01).The tumor size, density, TDR, SUVmax and MTV in the no lymph node me tastasis group were2.00(1.50～2.70)cm,7.0(-135.0～29.0)HU,0.52(0.30～0.92),3.90(1.80～9.10) and0.92(0.00～4.66)cm3respectively. In the lymph node metastasi s group, they were3.20(2.40～4.03)cm,32.0(26.0-38.0)HU,0.25(0.14～0.35),14.10(10.78～19.83) and15.33(4.77～26.98)cm3respectively. There were significant di fferences between the two groups (Z=-5.314,-5.878,-6.185,-7.025,-6.918, P<0.01).The differences in tumor size, density, TDR, SUVmax and MTV were obser ved when NO compared to the other N groups (P<0.05). Howerve, there were n o differences when N1～3groups were compared between any two groups(P<0.05). The lymph node size and SUVmax in the non-metastatic lymph node group were7.0(5.0～10.0)mm and2.4(2.0～4.4) respectively. In the metastatic lymph n ode group, they were13.0(10.0～16.5)mm and7.1(4.3～13.2) respectively. There were significant differences between the two groups(Z=-9.978,-10.419,P<0.01). The metastasis probability was extremely low when lymph node dentisy score was0or1(3%,2/59). On the contrary, all lymph node scored3were metastatic. Non-metastatic lymph nodes were scored0in term of the margin. On the contra ry, all lymph node scored1were metastatic. The most lymph nodes scored0(97%) in term of distribution were non-metastatic and most scored1(79%) were me tastatic. There were obvious differences in density, margin and distribution scores between the two groups (χ2=65.494,95.783,-10.419,P<0.01).Results from multi-viriables binary logistic regressionRegarding the patient gender, primary lung tumor imaging type and the prim ary tumor size, density, TDR, SUVmax and MTV, lymph node size, SUVmax, d ensity, margin and distribution scores as the independent variables, the lymph no de metastatic status as binary dependent variable, the results of logistics regressi on analysis showed the primary tumor SUVmax and lymph node size, density an d distribution score entered the regression equation (WALD=15.595,8.775,6.442,30.462,P<0.05).OR values were1.127,1.266,2.965and14.390respectively. The mathematical model based on18F-FDG PET/CT was stated by P=ey.(1+ey)y=-7.531+0.120×primary tumor SUVmax+0.236×lymph node size+1.087×lymph node density score+3.194×lymph node distribution score;P was expressed by the probability to lymph node metastasis; The unit of lymph node size is mm.Results from ROC curveThe ROC curves were obtained by lymph node size, lymph node SUVmax a nd mathematical model. The AUC of mathematical model, lymph node size and lymph node SUVmax was0.931,0.859and0.870. The differences of AUC were observedve (Z=2.057,2.006,P<0.05). According to maximum Youden index pri nciple, the cutoff values for lymph node size is10.5mm and4.05for lymph nod e SUVmax.ConclusionPatient age, primary lung cancer lesion location (deffrent lung lobe) and path ological types (adenocarcinoma and squamous cell carcinomas) cannot predict ly mph node metastasis in lung cancer. Male patients with lung cancer than female patients, the central type lung cancer than peripheral type develop easierly lymp h node metastasis.The larger primary tumor size, higher tumor density, smaller TDR and higer SUVmax and MTV mean the higer probability to lymph node metastasis in lun g cancer. But they are all not relavant to N1～3stage.The larger lymph node size and higher SUVmax mean higher probality to ly mph node metastasis. The probality to lymph node metastasis is extremely lower when calcification mottling or fatty is observed in lymph node. Unclear margin and necrosis are characteristic signs for the metastatic lymph node. When the ly mph nodes with increased radiactivy uptake distribute asymmetrically, they are m uch more likely metastatic.With the combination of both primary lesion features and lymph node featur es in18F-FDG PET/CT, the mathematical model to prediction lymph node metast sis in lung cancer has a higher diagnostic efficiency than lymph node size or ly mph node SUVmax. Part IIIThe Predictive Value of18F-FDG PET/CT on the Invasion of cTl Stage Pulmonary AdenocarcinomaBackgroud Along with the popularity of chest MSCT in clinical application, the incidence of cTl stage lung cancer increases rapidly, with about60%for a denocarcinoma. Different histological types of pulmonary adenocarcinoma are obv iouse different in biological behaviour. It is most important to evaluate the invasi on of cTl stage pulmonary adenocarcinoma for patient’s prognosis and individual ized treatment. Chest CT can provide anatomic details in tumors and PET can s how the metabolic features. The combination of CT and PET can provicd compr ehensive imageing information and maybe provide objective imaging evidence to evaluate the invasion of cTl stage pulmonary adenocarcinoma.Purpose To evaluate the predictive value of18F-FDG PET/CT scan on the i nvasion of cTl stage pulmonary adenocarcinoma by analyzing the relationship be tween their features in18F-FDG PET/CT and results of the pathological examinat ion retrospectively.Meterials and Methods Fifty-five patients with cTl stage pulmonary adenoc arcinoma underwent primary leision resection with regional lymph node dissectio n and18F-FDG PET/CT scan before the operation from June2010to August2013. The pathological results showed that:4patiets with mediastinal lymph node metastasis,11with intratumoral vessel invasion and6with both mediastinal lym ph node metastasis and intratumoral vessel invasion, namely LN/IVI(+) group(n=21);22patients with well differentiation,20patients with moderate differentiation and13with poor differentiation;31patients with pleural invasion, namely PI(+) group;16patients divided into the nipple in the priority group and25into the gland bubble in the priority group in terms of their histological classification, th e others divided into other histological types. The maximum long diameter length and short diameter length crossing a tumor section in CT lung window and me diastinal window were measured respectively. The maximum long diameter length was regarded as the size of tumor. TDR (Tumor Disappearance Rate) was calcu lated by the formula,1-long diameter lengthxshort diameter length in CT media stinal window/long diameter lengthx×short diameter length in CT lung window. I n Siemens syngo station, by using True D software to draw3D ROI, the maxiu m standarded uptake value of pixels in ROI was regarded as tumor SUVmax. Pa thology specimens after HE staining were observed for the histopathologic type, differentiation degree, intratumoral vessel invasion and the pleura invasion and ly mph node metastasis. Normal distribution measurement data were experessed with mean^standard deviation, Non-normal distribution measurement data were expere ssed with the median and quartiles, M (Q1～Q3). The differences in primary tu mor size, TDR and SUVmax between the LN/IVI(-) group and LN/IVI(+) group, well differentiation, moderate differentiation and poor differentiation groups, PI(-) group and PI(+) group, the nipple in the priority group and gland bubble in the priority group, were tested by Mann-Whitney test respectively. The relationships between the differentiation degree and primary tumor size, TDR and SUVmax were analysed by spersman correlation analysis. According to the principle of lar gest Youden index, by using Receiver Operating Characteristic curve, the cutoff values of primary tumor TDR and SUVmax to predict the mediastinal lymph no de metastasis and intratumoral vessel invasion were determined.ResultsRelationship between the imaging features and mediastinal lymph node metastasis and intratumoral vessel invasionThe primary tumor size was1.50(1.15～2.50) cm, TDR was0.73(0.40～1.00) and SUVmax was2.10(1.28～5.10) in the LN/IVI(-) group. The primary tumor size was2.20(1.50～2.95)cm, TDR was0.30(0.20～0.48) and SUVmax was10.70(4.40～15.00) in the LN/IVI(+) group. The differences in primary tumor size was not statistical significant(P>0.05)but not for TDR and SUVmax (P<0.01). The AUC obtained by ROC curve was0.814for primary tumor TDR and0.816for SUVmax. There was no significant difference between them (P>0.05). When the cutoff value was0.495for TDR,3.25for SUVmax, the sensitivity was71%,86%and specificity was81%,65%in predicting the mediastinal lymph node metastasis and intratumoral vessel invasion。Relationship between the imaging features and differentiationThe primary tumor size was1.50(1.00～2.05) cm, TDR was0.94(0.67～1.00) and SUVmax was1.55(1.00～2.60) in the well differentiation group. They were2.30(1.50～2.73)cm,0.39(0.28～0.59) and5.60(3.50～10.85) in the moderate differentiation group. They were1.80(1.50～3.00)cmã€0.27(0.20～0.42) and12.70(9.30～18.75) in the poor differentiation group. There were significant statistical differences in the primary tumor size, TDR and SUVmax between the well differentiation group and moderate differentiation group, so did between the well differentiation groups and poor differntation groups(P<0.05). Howerve, the unique difference in SUVmax was observed between the moderate differnatation and poor differentiation group (P<0.05). The spearman correlation analysis showed a low garde correlation (r=0.326,P<0.05) between primary tumor size and differentiation degree, but a high grade negative correlation (r=-0.700,P<0.01) between TDR and the differentiation degree, a high grade positive correlation (r=-0.773.P<0.01) between SUVmax and the differentiation degree.Relationship between the imaging features and the pleura invasionThe primary tumor size was1.50(1.00～2.35) cm, TDR was0.65(0.35～1.00) and SUVmax was3.20(1.33～6.30) in the PI(-) group. They are2.00(1.50-2.80) cm,0.46(0.27～0.83) and5.20(2.60～10.90) in PI(+) group. There were statistical differences in the primary tumor size and SUVmax between the two groups (P<0.05), but not for primary tumor TDR (P>0.05). The AUC of primary tumor size and SUVmax was0.674and0.656respectively.Relationship between the imaging features and the histological typeThe primary tumor size was1.60(1.20～2.85) cm, TDR was0.48(0.27～0.96) and SUVmax was3.05(1.43～9.83) in the the nipple in the priority group. They are1.80(1.10～2.50) cm,0.60(0.39～0.84) and4.40(2.10～11.20) in the gland bubble in the priority group. The differences in them were all not statistical significant (P>0.05).ConclusionThe cTl stage pulmonary adenocarcinoma with the samller TDR and higer SUVmax is be more likely to develop mediastinal lymph node metastasis and intratumoral vessel invasion. The primary tumor TDR and SUVmax have predictive value for the mediastinal lymph node metastasis and intratumoral vessel invasion in cTl stage pulmonary adenocarcinoma. The cTl stage pulmonary adenocarcinoma with the samller TDR and higer SUVmax is tendentious to be poor differentiation. The primary tumor TDR and SUVmax have predictive value for the defferntiation degree in cTl stage pulmonary adenocarcinoma.The cTl stage pulmonary adenocarcinoma with the larger tumor size and higer SUVmax is likely to develop pleural invasion, but the predictive value is limited.The primary tumor size, TDR and SUVmax maybe can not reflect the histological type of cTl stage pulmonary adenocarcinoma. |
PDF Full Text Request