Solitary Ground-glass Opacity Nodules Of Pulmonary Adenocarcinoma:Combination Of ¹⁸F-FDG PET/CT And High-resolution Computed Tomography Features To Predict Invasive Adenocarcinoma

Objective:To research the correlation between the invasion degree of stage IA pulmonary adenocarcinoma patients with solitary ground-glass opacity nodules and the imaging features of¹⁸F-FDG PET/CT and high resolution CT?HRCT?.To explore the feasibility of¹⁸F-FDG PET/CT combined with HRCT in predicting the invasion degree of IA stage lung adenocarcinoma.Method:This retrospective study enrolled 52 consecutive stage IA pulmonary adenocarcinoma patients with solitary ground-glass opacity nodules.According to the degree of invasion,it can be divided into two pathological types:minimally invasive adenocarcinoma?MIA?and invasive adenocarcinoma?IAC?.The characteristics and measurements of the ground-glass opacity nodules in the minimally invasive adenocarcinoma and invasive adenocarcinoma groups on Positron Emission Tomography/Computed Tomography and high-resolution CT were compared and analyzed.Consecutive fifty-two patients with a mean age of 57 years±8.4 were included our study,with no anti-tumoral therapies,no direct evidence of synchronous primary or prior malignancy in the past 5 years.within one week interval of all PET/CT,HRCT,and GGN resection,All patients fasted for at least 6 hours prior to PET/CT study.Insulin was discontinued at least 6 hours before examination,and a serum blood glucose level was verified to be below 10 mmol/L.All patients received FDG intravenously from 333 to 481 MBq,and then rested quietly around 60minutes.Low-dose registration CT and a wholebody PET were acquired from head to mid-thigh.For a semi-quantitative analysis of FDG uptake,a large oval volume of interest?VOI?was used.All patients underwent a breath-hold HRCT scan immediately after routine PET/CT scanning at our hospital.Two radiologists assessed the CT feature of GGN in position?subpleural/perifissural and parenchymal?,type?pure GGN and mixed GGN?,shape?round/oval and polygonal/irregular?,margin?smooth and lobulated/spiculated?,internal characteristics?bronchus sign and cystic appearance?,and adjacent structures?pleural indentation and vascular convergence?.Bronchus sign was classified into natural,dilated and distorted,and cut-off.Cystic appearance was defined as an oval,round,or large area of low attenuation within GGN.The diameter of GGN(D_GGN)was as the longest diameter on the transverse standard lung window image.For a mixed GGN,the diameter of the solid component(D_solid)was also measured in the longest diameter as the D_GGNGN on the transverse standard lung window image.The attenuation value of the GGO component(CT_GGO)and normal lung parenchyma adjacent to GGO(CT_LP)were measured using an oval region of interest covering the identical objects.The difference of CT_GGOGO and CT_LP(?CT_GGO–LP)was calculated as follows:CT_GGO–CT_LP.To determine the differences of HRCT features of the GGN between MIA and IAC,we performed the Chi-square test or for qualitative data and the student t test or Mann-Whitney U test for quantitative data.We also used ROC analysis and Youden index to identify the feasible preliminary threshold values of SUV_maxax at PET/CT and CTR at HRCT for distinguishing invasive adenocarcinoma from minimally invasive adenocarcinoma in stage IA patients with solitary GGN.Subsequently,binary logistic regression analysis was employed to exclude potential confounders of other HRCT features with statistical significant difference.Then,a logistic regression equation including SUV_max,CTR,and other HRCT features was formed by multivariate logistic regression analysis to compute 52 exact probability values,which were used to determine an optimal probability value by ROC analysis and Youden index.According to the optimal probability value,the ultimate combined optimal cutoff values of the SUV_max,CTR,and other HRCT features were identifiedto predict invasion of stage IA lung adenocarcinoma with GGO component.Finally,the sensitivity,specificity,positive predictive value,negative predictive value,and accuracy among the preliminary threshold values of SUV_maxax or CTR,and the ultimate combined optimal cutoff values of the SUV_max,CTR,and other HRCT features were compared via the McNemar test to differentiate IAC from MIA in stage IA lung adenocarcinoma patients with solitary GGN.Statistical analyses were performed by using software?SPSS version 19.0?.All variables with a value of p<0.05 were considered to indicate statistically significant differences.Results:The morphological characteristics and CT measurements of HRCT between the two groups were analyzed statistically.IAC cases preferentially manifested as mixed GGN?p<0.05?and cystic appearance was more common in IAC GGN?p<0.05?.Significant differences were also found in the D_GGN,D_solid,and CTR between the two groups?14.0mm vs 20.5mm,2.1mm vs 8.8mm,14.4%vs 41.9%,respectively??Feasible preliminary threshold values for CTR and SUV_maxax were 0.35and 1.04,respectively?While areas under the ROC curve were 0.835 for CTR?p<0.05?and 0.803 for SUV_max?p<0.05?.According to the two cutoff values,CTR and SUV_maxax were performed respectively into binary data,which were used to control confounders of other HRCT features with statistically significant difference by binary logistic regression analysis.nodule type was the only independent HRCT characteristic parameter.The corresponding multivariate logistic regression equation was formed.The exact probability values of 52 cases were calculated by backtesting the above-mentioned multivariate logistic regression equation with the specific evaluations of the SUV_max,CTR,and Nodule Type.Afterwards,the 52 exact probability values were analyzed by ROC analyses and Youden index to determine the ultimate combined optimal cutoff values of CTR>0.42,SUV_max>1.57,and mixed GGN of Nodule Type.In terms of all the performances,the ultimate combined optimal cutoff values of CTR>0.35,SUV_max>1.04,and mixed GGN demonstrated a higher sensitivity of 86.8%,a favourable negative predictive value of 76.9%and accuracy of 86.5%,but a balanced specificity and positive predictive value.The sensitivity of the combined CTR>0.42,SUV_max>1.57 with mixed GGN was higher than that of SUV_max>1.04 alone or CTR>0.35 alone via the McNemar test?both p<0.05?.Conclusion:1.combination of SUV_max?CTR and nodule type can predicting invasion degree of stage IA pulmonary adenocarcinoma patients with solitary ground-glass opacity nodules2.combination of CTR>0.42,SUVmax>1.57 and mixed GGN had a sensitivity of 86.8%?negative predictive value of 76.9%and accuracy of 86.5%in predicting the invasiveness of IA lung adenocarcinoma with solitary ground-glass opacity nodules.