Individual data meta-analysis for the study of survival after pulmonary metastasectomy in colorectal cancer patients: A history of resected liver metastases worsens the prognosis

Objectives: To assess the impact of a history of liver metastases on survival in patients undergoing surgery for lung metastases from colorectal carcinoma. Methods: We reviewed recent studies identified by searching MEDLINE and EMBASE using the Ovid interface, with the following search terms: lung metastasectomy, pulmonary metastasectomy, lung metastases and lung metastasis, supplemented by manual searching. Inclusion criteria were that the research concerned patients with lung metastases from colorectal cancer undergoing surgery with curative intent, and had been published between 2007 and 2014. Exclusion criteria were that the paper was a review, concerned surgical techniques themselves (without follow-up), and included patients treated non-surgically. Using Stata 14, we performed aggregate data and individual data meta-analysis using random-effect and Cox multilevel models respectively. Results: We collected data on 3501 patients from 17 studies. The overall median survival was 43 months. In aggregate data meta-analysis, the hazard ratio for patients with previous liver metastases was 1.19 (95% CI 0.90e1.47), with low heterogeneity (I2 4.3%). In individual data meta-analysis, the hazard ratio for these patients was 1.37 (95% CI 1.14e1.64; p < 0.001). Multivariate analysis identified the following factors significantly affecting survival: tumour-infiltrated pulmonary lymph nodes (p < 0.001), type of resection (p ¼ 0.005), margins (p < 0.001), carcinoembryonic antigen levels (p < 0.001), and number and size of lung metastases (both p < 0.001). Conclusions: A history of liver metastases is a negative prognostic factor for survival in patients with lung metastases fromcolorectal cancer.We registered themeta-analysis protocol in PROSPERO(CRD42015017838).