Clinicoradiographic Predictors of Aggressive Behavior in Resected Malignant Ground Glass Opacity Lesions
David Nelson1, Myrna Godoy1, Marcelo Benveniste1, Jitesh Shewale1, Arlene Correa1, Jonathan Spicer2, Wayne Hofstetter1, Reza Mehran1, David Rice1, Boris Sepesi1, Garett Walsh1, Ara Vaporciyan1, Stephen Swisher1, Jack Roth1, Mara Antonoff1
1MD Anderson Cancer Center, Houston, TX;2McGill University Health Center Research Institute, Montreal, QC, Canada
OBJECTIVES: The unclear prognostic significance of ground glass opacities (GGOs) continues to pose a diagnostic challenge for thoracic surgeons, and reliable predictors of aggressive biologic behavior are in need. In this study, we aimed to evaluate the predictive role of clinicoradiographic features on the tumor grade and pathologic stage of resected malignant GGO lesions.
METHODS: A retrospective review was performed of patients who had NSCLC lesion described as a GGO on a preoperative imaging report and underwent surgical resection from 2008-2013. Inclusion criteria were as follows: T1-4, N0-1, and no receipt of induction chemotherapy. Pre-operative images were reviewed and detailed radiographic elements were collected and supplemented with clinical data from chart review. Clinicoradiologic data were then evaluated for correlation with post-operative pathologic findings.
RESULTS: 222 patients met inclusion criteria. Detailed radiographic analyses localized the malignancy within the known GGO for 67 patients, who thus comprised the final cohort. Further analyses were performed with these individuals. Lesion location by laterality, peripheral vs. central field, pure GGO vs. partial solid nodule, heterogeneity, cystic appearance, pleural tags, spiculation, reticulation, and lobulation were not associated with higher grade nor pathologic stage. On univariate logistic regression, lesion size was associated with tumor grade (p=0.012). Air bronchogram displayed a nonsignificant trend toward association with tumor grade (p=0.103). However, only lesion size was associated with tumor grade on multivariate regression (p=0.025). For pathologic stage, lesion size and presence of air-bronchograms also displayed nonsignificant trends toward association with stage II or greater (p=0.07 and p=0.134, respectively). Among clinical factors, diabetes and smoking were associated with stage II or greater (p=0.015 and p=0.047, respectively). Multivariate regression of clinicoradiographic features showed association of diabetes and smoking history with pathologic stage (p=0.009 and p=0.024, respectively).
CONCLUSIONS: For patients with resected NSCLC arising in GGO, lesion size is associated with tumor grade, while diabetes and smoking history were associated with more advanced pathologic stage. These clinicoradiographic findings may help provide prognostic information regarding malignant GGOs, and may ultimately guide operative therapy in terms of extent of resection.
|OR||(95% confidence interval)||p-value|
|Size||1.03||(1.0 - 1.08)||p=0.07|
|Type (pure GGO vs. PSN)||0.58||(0.16 - 2.08)||p=0.40|
|Heterogenous||2.05||(0.40 - 10.4)||p=0.39|
|Air-Bronchogram||2.65||(0.74 - 9.4)||p=0.13|
|Cystic||1.58||(0.15 - 16.6)||p=0.71|
|Pleural Tag||1.79||(0.46 - 6.98)||p=0.40|
|Spiculation||1.58||(0.15 - 16.6)||p=0.71|
|Reticulation||1.03||(0.27 - 3.87)||p=0.97|
|Lobulation||1.33||(0.35 - 5.09)||p=0.67|
Back to 2017 Program