Patient data for those diagnosed with metastatic cancer between 1992 and 2019 were extracted fromt he National Cancer Institute's Surveillance, Epidemiology, and End Results (SEER) program (SEER 18 database). The SEER program is a network of population-based incident tumor registries from various regions in the United States, representing 28% of the country's population, and includes information on incidence, survival, and treatment (such as radiation therapy, surgery, and chemotherapy).
The study population was from geographically distinct US regions, chosen to represent the racial and ethnic heterogeneity of the country. The SEER program has evolved since its inception in the United States in 1973. As of 2017, SEER has up to 36 years of longitudinal and ongoing data collection, with a representative sample size of more than 6 million cancer cases, and a comprehensive quality assurance process.
To generate novel nomograms, Fine-Gray sub-distribution hazard regression models were fit based on clinical and demographic data, such as primary cancer type, metastasis to bone, brain, liver, and/or lung, race, and age, similar to a recent pan-cancer analysis of metastatic phenotypes. The training set was derived from the SEER 2010-2012 data, while the SEER 2013-2015 data was used for validation. These time frames were chosen to ensure no overlap between patients in the two datasets. The models were internally validated using Uno's concordance (C) index, where a higher C-index indicates better discrimination. These results were then used to create nomograms to predict survival in patients living with metastatic cancer for the initial diagnosed metastatic cancer and other causes of death at 1, 3, and 5 years after diagnosis.