Prognostic model for survival in patients with neuroendocrine carcinomas of the cervix: SEER database analysis and a single-center retrospective study

by Caixian Yu, Xiaoliu Wu, Shao Zhang, Lan Zhang, Hongping Zhang, Hongying Yang, Min Zhao, Zheng Li

Objective

Neuroendocrine carcinoma of the cervix (NECC) is extremely rare in clinical practice. This study aimed to methodologically analyze the clinicopathological factors associated with NECC patients and to develop a validated survival prediction model.

Methods

A total of 535 patients diagnosed with NECC between 2004 and 2016 were identified from the Surveillance, Epidemiology and End Results (SEER) database, while 122 patients diagnosed with NECC at Yunnan Cancer Hospital (YCH) from 2006 to 2019 were also recruited. Patients from the SEER database were divided into a training cohort (n = 376) and a validation cohort (n = 159) in a 7:3 ratio for the construction and internal validation of the nomogram. External validation was performed in a cohort at YCH. The Kaplan-Meier method was used for survival analysis, the Log-rank method test was used for univariate analysis of prognostic influences, and the Cox regression model was used for multivariate analysis.

Results

The 3-year and 5-year overall survival (OS) rates for patients with NECC in SEER were 43.6% and 39.7%, respectively. In the training cohort, multivariate analysis showed independent prognostic factors for NECC patients including race, tumor size, distant metastasis, stage, and chemotherapy (p<0.05). For extended application in other cohorts, a nomogram including four factors without race was subsequently created. The consistency index (C-index) of the nomogram predicting survival was 0.736, which was well-validated in the validation cohorts (0.746 for the internal validation cohort and 0.765 for the external validation cohort). In both the training and validation cohorts, the 3-year survival rates predicted by the nomogram were comparable to the actual ones. We then succeeded in dividing patients with NECC into high- and low-risk groups concerning OS using the nomogram we developed. Besides, univariate analysis showed that chemotherapy ≥4 cycles may improve the OS of patients at YCH with NECC.

Conclusion

We successfully constructed a nomogram that precisely predicts the OS for patients with NECC based on the SEER database and a large single-center retrospective cohort. The visualized and practical model can distinguish high-risk patients for recurrence and death who may benefit from clinical trials of boost therapy effectively. We also found that patients who received more than 4 cycles of chemotherapy acquired survival benefits than those who received less than 4 cycles.