| Part One Clinical and Imaging features of Primary and Secondary Pulmonary Lymphoma: Discriminant Analysis Based on Logistic Regression MethodObjective: To establish a predictive model for quantitative assessment of primary and secondary pulmonary lymphoma and to compare the characteristics between groups.Materials and methods: A total of 111 patients with pulmonary lymphoma diagnosed from 2004 to 2018 were retrospectively reviewed.There were 24 males and 26 females with median age 56 years in 50 cases primary pulmonary lymphoma(PPL)patients,and 40 males and 21 females with median age 59 years in 61 cases secondary pulmonary lymphoma(SPL)patients.The clinical features included symptoms(asymptomatic,B symptoms and respiratory symptoms),serum lactate dehydrogenase(LDH)level and stage.Two radiologists evaluated the patterns and characteristics of CT findings.χ2 tests or Fisher’s exact tests were performed for the comparison of the differences in clinical features and imaging findings between groups.The significant variables were further applied to discrimination between PPL and SPL based on Logistic regression analysis.p value on the left side of the inflection point of the ROC curve according to discrimination probability was used as a criterion to discriminate PPL and SPL.Results: Logistic discriminant function for prediction of PPL and SPL was Logit p=-0.746-0.125 B symptoms(without=0,with=1)-0.502 LDH level(normal=0,elevated=1)-2.744 stage(I+II=0,III+IV=1)+0.296number(single=0,multiple=1)+1.294 air bronchogram(without=0,with=1)-1.618 halo sign(without=0,with=1)+1.669 bulging fissure sign(without=0,with=1)+2.062 bronchiectasis(without=0,with=1)-0.868 pleural involvement(without=0,with=1)-0.347 hilar/mediastinal lymphadenopathy(without=0,with=1).p=0.575 was used as a criterion.The false positive rate was 12.6%,the sensitivity was 88.5%,and the specificity was 86.0%.In this sample,the rations of the number of SPL patients who were staged I+II and III+IV,and without and with air bronchogram,halo sign and bronchiectasis were 0.064,and 3.646,0.198 and 7.860,respectively.Conclusions: Patients staged III+IV and who show halo sign suggest the diagnosis of SPL.Air bronchogram with bronchiectasis is a characteristic CT finding of PPL patients.Discriminant Analysis based on Logistic regression method can be used as a prediction method of PPL and SPL,which has a high clinical value.Part Two Patients with pulmonary B-cell non-Hodgkin’s lymphoma: Characteristic CT Findings and Role in Predicting the SurvivalObjective: To evaluate the value of characteristic CT findings to predict the survival of patients with pulmonary B-cell non-Hodgkin’s lymphoma(NHL).Materials and methods: We retrospectively enrolled 84 patients with pulmonary B-cell NHL confirmed by pathology between 2004 and 2018.All patients underwent chest CT scan at initial diagnosis in our hospital.Characteristic CT findings and clinicopathological features of them were analyzed.Cox regression models were performed to analyze the relation between characteristic CT findings and OS(overall survival)and PFS(progression-free survival).Results: Air bronchogram occurred more frequently in patients with early stage,PPL and indolent histological type than those with advanced stage,SPL and aggressive histological type(all p<0.05).Halo sign was observed more in the SPL group(p=0.004).While the presence rate of cross lobe sign was higher in patients with PPL(p=0.010).Pleural involvement and hilar/mediastinal lymphadenopathy were observed more in patients with SPL and aggressive histological type(all p<0.05).In survival analyses,the number of lung lesion,cross lobe sign and pleural involvement were independent prognostic factors for PFS;while halo sign and pleural involvement were significantly correlated with OS(all p<0.05).Besides,more aggressive,advanced stage and male patients showed worse outcomes.Conclusions: Halo sign and pleural involvement are independent prognostic factors for OS;while the number of lung lesion,cross lobe sign and pleural involvement are correlated with PFS. |
PDF Full Text Request