OBJECTIVES: The purpose of this study was to use Hounsfield unit (HU) thresholds of computed tomography (CT) images to predict pathological lymph node metastasis and tumour invasiveness of cT1N0M0 lung adenocarcinoma on 3D evaluations. METHODS: Preoperative CT images of 211 lesions of surgically resected cT1N0M0 lung adenocarcinoma were retrospectively examined. The tumour size was calculated in 1D, 2D and 3D views. Tumours with -300 HU and over were defined as 'solid tumours', and those between -800 and -301 HU were defined as 'ground glass opacity tumours'. Tumours with -800 HU and over were assumed to be the whole tumour entity. The proportion of 'solid tumour' within the whole tumour entity was also calculated as the 'solid tumour ratio'. These were compared with pathological information. RESULTS: Solid tumour size and ratio were positively correlated with microscopic invasion to pleura, vessels and lymphatics in all dimensional evaluations. Pathological lymph node metastases were also well predicted by solid tumour size and ratio in all dimensional evaluations. The P-values for the receiver operating characteristic (ROC) curves of 1D, 1D ×2, 2D and 3D evaluations were: solid tumour size P = 0.013, 0.014 and 0.032; and solid tumour ratio 0.016, 0.0032 and <0.0001. In comparisons of 1D, 2D and 3D evaluations, 'solid tumour size' of the area under the curve (AUC) of ROC to detect pathological lymph node metastases was not significant. However, strikingly, the 3D solid tumour ratio was found to be significantly more accurate for the prediction of pathological lymph node metastases than the 1D and 2D solid tumour ratios on ROC evaluation (AUC: 1D 0.736, 2D 0.803 and 3D 0.882; P-values for the AUC comparisons were P = 0.013 for 3D versus 1D and P = 0.022 for 3D versus 2D). The correlations of subtypes of adenocarcinoma and the 3D solid tumour ratio were also investigated. Subtypes of adenocarcinoma were well correlated with the 3D solid tumour ratio. CONCLUSIONS: Preoperative 3D CT using threshold values of -800 and -300 HU was useful for predicting pathological lymph node metastases and tumour invasiveness of cT1N0M0 lung adenocarcinoma.
OBJECTIVES: The purpose of this study was to use Hounsfield unit (HU) thresholds of computed tomography (CT) images to predict pathological lymph node metastasis and tumour invasiveness of cT1N0M0 lung adenocarcinoma on 3D evaluations. METHODS: Preoperative CT images of 211 lesions of surgically resected cT1N0M0 lung adenocarcinoma were retrospectively examined. The tumour size was calculated in 1D, 2D and 3D views. Tumours with -300 HU and over were defined as 'solid tumours', and those between -800 and -301 HU were defined as 'ground glass opacity tumours'. Tumours with -800 HU and over were assumed to be the whole tumour entity. The proportion of 'solid tumour' within the whole tumour entity was also calculated as the 'solid tumour ratio'. These were compared with pathological information. RESULTS: Solid tumour size and ratio were positively correlated with microscopic invasion to pleura, vessels and lymphatics in all dimensional evaluations. Pathological lymph node metastases were also well predicted by solid tumour size and ratio in all dimensional evaluations. The P-values for the receiver operating characteristic (ROC) curves of 1D, 1D ×2, 2D and 3D evaluations were: solid tumour size P = 0.013, 0.014 and 0.032; and solid tumour ratio 0.016, 0.0032 and <0.0001. In comparisons of 1D, 2D and 3D evaluations, 'solid tumour size' of the area under the curve (AUC) of ROC to detect pathological lymph node metastases was not significant. However, strikingly, the 3D solid tumour ratio was found to be significantly more accurate for the prediction of pathological lymph node metastases than the 1D and 2D solid tumour ratios on ROC evaluation (AUC: 1D 0.736, 2D 0.803 and 3D 0.882; P-values for the AUC comparisons were P = 0.013 for 3D versus 1D and P = 0.022 for 3D versus 2D). The correlations of subtypes of adenocarcinoma and the 3D solid tumour ratio were also investigated. Subtypes of adenocarcinoma were well correlated with the 3D solid tumour ratio. CONCLUSIONS: Preoperative 3D CT using threshold values of -800 and -300 HU was useful for predicting pathological lymph node metastases and tumour invasiveness of cT1N0M0 lung adenocarcinoma.
Authors: Nanda Horeweg; Carlijn M van der Aalst; Erik Thunnissen; Kristiaan Nackaerts; Carla Weenink; Harry J M Groen; Jan-Willem J Lammers; Joachim G Aerts; Ernst T Scholten; Joost van Rosmalen; Willem Mali; Matthijs Oudkerk; Harry J de Koning Journal: Am J Respir Crit Care Med Date: 2013-04-15 Impact factor: 21.405