Competing-risks nomograms for predicting cause-specific mortality in parotid-gland carcinoma: A population-based...

Parotid-gland carcinoma (PGC) is a relatively rare tumor that comprises a group of heterogeneous histologic subtypes. We used the Surveillance, Epidemiology, and End Results (SEER) program database to apply a competing-risks analysis to PGC patients, and then established and validated predictive nomograms for PGC.

Specific screening criteria were applied to identify PGC patients and extract their clinical and other characteristics from the SEER database. We used the cumulative incidence function to estimate the cumulative incidence rates of PGC-specific death (GCD) and other cause-specific death (OCD), and tested for differences between groups using Gray's test. We then identified independent prognostic factors by applying the Fine-Gray proportional subdistribution hazard approach, and constructed predictive nomograms based on the results. Calibration curves and the concordance index (C-index) were employed to validate the nomograms.

We finally identified 4,075 eligible PGC patients who had been added to the SEER database from 2004 to 2015. Their 1-, 3-, and 5-year cumulative incidence rates of GCD were 10.1%, 21.6%, and 25.7%, respectively, while those of OCD were 2.9%, 6.6%, and 9.0%. Age, race, World Health Organization histologic risk classification, differentiation grade, American Joint Committee on Cancer (AJCC) T stage, AJCC N stage, AJCC M stage, and RS (radiotherapy and surgery status) were independent predictors of GCD, while those of OCD were age, sex, marital status, AJCC T stage, AJCC M stage, and RS. These factors were integrated for constructing predictive nomograms. The results for calibration curves and the C-index suggested that the nomograms were well calibrated and had good discrimination ability.

We have used the SEER database to establish-to the best of our knowledge-the first competing-risks nomograms for predicting the 1-, 3-, and 5-year cause-specific mortality in PGC. The nomograms showed relatively good performance and can be used in clinical practice to assist clinicians in individualized treatment decision-making.