OBJECTIVES This study was conducted to verify the optimal extent of lymph node dissection or sampling during lung cancer surgery based on the sentinel node (SN) map created by computed tomography (CT) lymphography. METHODS From April 2010 to January 2015, patients with clinical stage I non-small-cell lung cancer, who were candidates for lobectomy or segmentectomy with standard hilar and mediastinal lymph node dissection, and in whom bronchus reached the tumour, were enrolled. An ultrathin bronchoscope was inserted to the target bronchus under the guidance of virtual bronchoscopic navigation images. CT images of the chest were obtained 30 s after 2.5 ml of iopamidol was injected. SNs were identified when the maximum CT attenuation value of the lymph nodes on postcontrast CT images increased by 30 Hounsfield units or more compared with the precontrast images. Patients underwent lobectomy with standard lymph node dissection. RESULTS SNs were identified in 36 (87.8%) of the 41 patients. The average number of SNs was 1.6 (range, 1–4). There was 1 false negative case; therefore, the accuracy of SN identification was 97.2% (35/36). In 5 (13.9%) of 36 patients, SNs were outside the lobe-specific lymph node station range (#11i from right S1, #7 from right S1, #4R from right S8, #12u from right S8, #7 and #12l from left S1 + 2). CONCLUSIONS CT lymphography demonstrated the diversity of lymphatic spreading patterns and there were cases in which lymph flows are found outside the lymph node dissection range.
