Preoperative CT features for prediction of ALK gene rearrangement in lung adenocarcinomas

Abstract

Aim: To identify preoperative features on computed tomography (CT) associated with ALK rearrangement in lung adenocarcinomas presenting as a nodule.

Materials and methods: This retrospective analysis included 56 patients with ALK rearrangement and 57 that were ALK-negative. All patients had surgically resected lung adenocarcinomas <3 cm. Univariate and multivariate analyses were conducted to analyse clinicopathological and CT features associated with ALK rearrangement. Receiver operating characteristic (ROC) analyses were performed to quantify the performance status of the model.

Results: ALK rearrangement was associated with lymph node metastases (p=0.004), later pathological stage (p=0.005), lower lobe (p=0.019), lobulation (p=0.006), thickened adjacent bronchovascular bundles (p=0.006), homogeneous tumour (p=0.008), absence of ground-glass opacity (GGO; p<0.001), absence of air bronchogram (p=0.010), smaller relative enhancement (p=0.019), and larger short axis of the largest lymph node (p=0.012). Cavity larger than 1 cm was found in 3 ALK-positive tumours while not in ALK-negative tumours. Multivariate analysis revealed a single predictive model with an AUC of 0.794 that lobulation (OR=4.50, p=0.026), GGO (OR=0.19, p=0.003), and short axis of the largest lymph node (OR=12.49, p=0.047) were independent predictors of ALK rearrangement status.

Conclusions: This study identified a modestly predictive radiological model to identify ALK rearrangement in small lung adenocarcinomas.