Assessing The Proportion Of Lepidic Growth In Invasive Adenocarcinoma With Pure Ground Glass Nodule Using CT Characteristics : A Pilot Study

Objective:To assess the proportion of lepidic growth in invasive adenocarcinoma(IA)with pure ground glass nodule(pGGN)on thin-slice CT using its clinical data and imaging characteristics and look for the best imaging indicators to predict lepidic growth more than 50% and the correlation between the immunohistochemical indexes and the proportion of lepidic growth.Materials and methods:Patients with pGGN on thin-slice CT undergoing resection in the first affiliated hospital of Dalian medical university diagnosed invasive adenocarcinoma in the period January 2013-January 2018 were enrolled in this study.All 80 cases were single IA lesions.According to chronological order,all cases were divided into primary cohort(January 2015 to January 2018)and validation cohort(January 2013 to December 2014).Primary cohort(59 pGGNs): validation cohort(21 pGGNs)=2.81: 1.All the patients completed the thin-slice CT scan within one week before the operation and completed the operation in the department of thoracic surgery of our hospital.Clinical data(age,sex,smoking history),surgical methods and immunohistochemical indicators of all subjects were collected retrospectively.Imaging characteristics of pGGNs on thin-slice CT were observed and measured,including mean CT value of the largest plane(mCT-LP),mean CT value of the whole(mCT-W),diameter and imaging signs(shape,margin,burr,air bronchogram,vacuolar sign,pleural traction sign,thickening blood vessels,etc.).All postoperative pathological sections of the IA lesions were reviewed and analyzed by the pathologists.Proportion of lepidic growth was recorded and All cases were divided into lepidic growth more than 50% and less than 50%.SPSS for Windows,version17.0(Inc,Chicago,IL,USA).In the primary cohort,the correlation between clinical data,imaging characteristics,immunohistochemical indicators and the proportion of lepidic growth was analyzed by Spearman.The differences of clinical data,imaging characteristics and immunohistochemical indicators between lepidic growth more than 50% and less than 50% was compared using chi-square test test for counting data,Student's t test for the measurement data conforming to the normal distribution and the Mann-Whitney U test for the non-normal distribution.ROC curve was analyzed to evaluate lepidic growth more than 50% by imaging features.The cutoff point of the CT value from the primary cohort was used to predict the proportion of lepidic growth in the validation cohort and this result was compared with pathological result as a gold standard using McNemar's test.A two-sided p value was always computed and a difference was considered significant at p < 0.05.Results:1.Primary cohortThere was a negative correlation between mCT-LP(median-495.89 Hu,-725.33~-300.03Hu)and the lepidic growth as well as mCT-W(median-499.05 Hu,-718.75~-315.98Hu)(p=0.00,r=-0.58;p=0.00,r=-0.77).There was a negative correlation between p53 mutation rate(median 10%,0%~95%)and the proportion of lepidic growth(p=0.02,r=-0.30).The mCT-LP(median-599.23 Hu,-725.33~-335.36Hu)and mCT-W(median-605.61 Hu,-718.75~-499.05Hu)in lepidic growth more than 50% were significantly lower than lepidic growth less than 50%(median-458.37 Hu,-618.18~-300.03Hu;median-420.38 Hu,-627.00~-315.98 Hu,all p=0.00).Area under the curve(AUC)was 0.91 when lepidic growth more than 50% was determined by mCT-W.When cutoff value of mCT-W was-527.64 Hu,the sensitivity was 89.70%,the specificity was 90.00%,the positive predictive value was 82.61%,negative predictive value was 97.22% and the diagnostic accuracy was 91.52%.2.Validation cohortThe cutoff value(-527.64Hu)from the primary cohort was used to predict and verify lepidic growth more than 50% in the 21 lesions of the validation cohort.And this result predicted by CT value was compared with final pathology.The results of the two judgments are basically the same(p =1.00).Conclusion:The quantitative imaging indicators of thin-slice CT can predict the proportion of lepidic growth.mCT-W is an ideal index to determine the proportion of lepidic growth.The optimal cut-off value is-527.64 Hu,negative predictive value was 97.22% and the diagnostic accuracy was 91.52%.