Nomograms for Predicting Risk and Survival of Esophageal Cancer Lung Metastases: a SEER Analysis
Wenhui He,Youzhen Yu,Ziting Yan,Na Luo,Wenwen Yang,Fanfan Li,Hongying Jin,Yimei Zhang,Xiaoli Ma,Minjie Ma
DOI: https://doi.org/10.7150/jca.92389
2024-04-23
Abstract:Background: The overall survival rate is notably low for esophageal cancer patients with lung metastases (LM), presenting significant challenges in their treatment. Methods: Through the Surveillance, Epidemiology, and End Results (SEER) program, individuals diagnosed with esophageal cancer between 2010 and 2015 were enrolled. Based on whether esophageal cancer metastasized to the lungs, we used propensity score matching (PSM) to balance correlated variables. Propensity score matching was a critical step in our study that helped to minimize the impact of possible confounders on the study results. We balanced variables related to lung metastases using the PSM method to ensure more accurate comparisons between the study and control groups. Specifically, we performed PSM in the following steps. First, we performed a univariate logistic regression analysis to screen for variables associated with lung metastasis. For each patient, we calculated their propensity scores using a logistic regression model, taking into account several factors, including gender, T-stage, N-stage, surgical history, radiotherapy history, chemotherapy history, and bone/brain/liver metastases. We used a 1:1 matching ratio based on the propensity score to ensure more balanced baseline characteristics between the study and control groups after matching. After matching, we validated the balance of baseline characteristics to ensure that the effect of confounders was minimized. We used logistic regression to identify risk variables for LM, while Cox regression was used to find independent prognostic factors. We then created nomograms and assessed their accuracy using the calibration curve, receiver operating curves (ROC), and C index. Results: In the post-PSM cohort, individuals diagnosed with LM experienced a median overall survival (OS) of 5.0 months (95% confidence interval [CI] 4.3-5.7), which was significantly lower than those without LM (P<0.001). LM has been associated to sex, T stage, N stage, surgery, radiation, chemotherapy, and bone/brain/liver metastases. LM survival was affected by radiation, chemotherapy, and bone/liver metastases. The nomograms' predictive power was proved using the ROC curve, C-index, and validation curve. Conclusion: Patients with LM have a worse chance of surviving esophageal cancer. The nomograms can effectively predict the risk and prognosis of lung metastases from esophageal cancer.