Development and Validation of Web-Based Nomograms to Precisely Predict Conditional Risk of Site-Specific Recurrence for Patients With Completely Resected Non-small Cell Lung Cancer: A Multiinstitutional Study

Abstract

There is currently no consensus regarding the optimal postoperative follow-up strategy for patients with completely resected non-small cell lung cancer (NSCLC). We aimed to develop web-based nomograms to precisely predict site-specific postoperative recurrence in patients with NSCLC and to guide individual surveillance strategies including when to follow up and what diagnostic tests to perform. We investigated the pattern of recurrence in a series of 2,017 patients with NSCLC (squamous cell carcinoma and nonlepidic invasive adenocarcinoma) who underwent complete surgical resection at Fudan University Shanghai Cancer Center (development cohort), and developed web-based clinicopathologic prediction models for conditional risk of site-specific recurrence based on Cox regression. The variables used in the analysis included sex, age, smoking history, tumor size, tumor histology, lymphovascular invasion, visceral pleural invasion, and pathologic TNM stage. A separate cohort of 3,308 patients with NSCLC from Shanghai Chest Hospital was used for external validation. In the development cohort and the external validation cohort for the established nomograms to predict overall recurrence, thorax recurrence, abdomen recurrence, neck recurrence, brain recurrence, and bone recurrence, the C-statistics of Harrell etal were 0.743 and 0.748, 0.728 and 0.703, 0.760 and 0.749, 0.779 and 0.757, 0.787 and 0.784, and 0.777 and 0.739, respectively. The calibration plots showed optimal agreement between nomogram-predicted 3-year recurrence-free survival and actual 3-year recurrence-free survival. These user-friendly nomograms can precisely predict site-specific recurrence in patients with completely resected NSCLC, based on clinicopathologic features. They may help physicians to make individual postoperative follow-up plans.