Dynamic estimates of survival of patients with poorly differentiated thyroid carcinoma: a population-based study.

BACKGROUND

The real-time prognostic data of patients with poorly differentiated thyroid carcinoma (PDTC) after surviving for several years was unclear. This study aimed to employ a novel method to dynamically estimate survival for PDTC patients.

METHODS

A total of 913 patients diagnosed with PDTC between 2014 and 2015 from the Surveillance, Epidemiology, and End Results (SEER) database, was recruited in our study. Kaplan-Meier method was used to estimate the overall survival (OS). The conditional survival (CS) outcomes of PDTC were analyzed and CS rates were calculated using the formula CS(y/x) = OS(y+x)/OS(x), whereby CS(y/x) denotes the probability of a patient enduring an additional y years subsequent to surviving x years following the diagnosis of PDTC. The least absolute shrinkage and selection operator (LASSO) regression was employed to identify prognostic predicters and multivariate Cox regression was utilized to develop a CS-nomogram. Finally, the performance of this model was evaluated and validated.

RESULTS

Kaplan-Meier survival analysis unveiled patient outcomes demonstrating an OS rate of 83%, 75%, and 60% respectively at the end of 3, 5, and 10 years. The novel CS analysis highlighted a progressive enhancement in survival over time, with the 10-year cumulative survival rate progressively augmenting from its initiation of 60% to 66%, 69%, 73%, 77%, 81%, 83%, 88%, 93%, and finally 97% (after surviving for 1-9 years, respectively) each year. And then 11 (11/15) predictors including age at diagnosis, sex, histology type, SEER stage, T stage, N stage, M stage, tumor size, coexistence with other malignancy, radiotherapy and marital status, were selected by LASSO analysis under the condition of lambda.min. Multivariate Cox regression analysis further highlighted the significant impact of all these predictors on the OS of PDTC and we successfully established and validated a novel CS-nomogram for real-time and dynamic survival prediction.

CONCLUSIONS

This was the first study to analyze the CS pattern and demonstrate a gradual improvement in CS over time in long-term PDTC survivors. We then successfully developed and validated a novel CS-nomogram for individualized, dynamic, and real-time survival forecasting, empowering clinicians to adapt and refine the patient-tailored treatment strategy promptly with consideration of evolving risks.