Causal Model of Survival after Pulmonary Metastasectomy of Colorectal Cancer: A Nationwide Prospective Registry

Abstract

Background

Although numerous existing studies have analyzed the prognostic factors of patients who have had surgical intervention for lung metastases of colorectal carcinoma (CRC), many of the results obtained until now have been contradictory. As a consequence, there is no established consensus about which group of prognostic factors could have a greater value when considered together. Methods

This was a multicenter prospective cohort study that included all patients who underwent a first pulmonary metastasectomy of CRC, with radical intent, during a 2-year period (March 2008 to February 2010). The follow-up continued until March 2013, and an analysis of disease-specific survival (DSS), determined from the first pulmonary metastasectomy, was implemented. The selection of the best submodel was taken based on their coefficient of determination (R2) and how parsimonious they were depending on the number of variables included. Results

The series, consisting of 522 patients, presented the following survival rates: median, 54.9 months; 3-year DSS, 69.4% (95% confidence interval [CI], 65% to 73.8%); and 5-year DSS, 46.1% (95% CI, 38.5% to 53.7%). The resulting survival model consisted of disease-free interval of 12 months or less (hazard ratio [HR], 1.76; 95% CI, 1.21 to 2.54; p = 0.003), carcinoembryonic antigen level exceeding 5 ng/mL (HR, 1.50; 95% CI, 1.04 to 2.17; p = 0.028), bilateral lung disease (HR, 1.81; 95% CI, 1.20 to 2.75; p = 0.005), and thoracic lymph node involvement (HR, 2.71; 95% CI, 1.44 to 5.12; p = 0.002). Conclusions

According to these results from the Spanish Group of Lung Metastases of Colo-Rectal Cancer, the combination of these four variables—disease-free interval, carcinoembryonic antigen level, laterality, and thoracic lymph node involvement—constitutes the first-choice survival causal model based on the clinical and pathologic factors most frequently referenced in literature.