Nomograms Constructed for Predicting Diagnosis and Prognosis in Cervical Cancer Patients with Second Primary Malignancies: A SEER Database Analysis

Abstract

Abstract Purpose Cervical cancer (CC) patients are more likely to develop second primary malignancies (SPMs) than general population. With the advancement in cancer therapy, CC patients are achieving long-term survival, leading SPMs to our attention. Our study aims to establish diagnostic and prognostic nomograms for CC patients with second primary malignancies (CCSPMs) to help make personalized follow-up plans and treatments. Methods Data of CCSPMs between 2000 and 2019 was extracted from SEER. The proportions and the average interval time of CCSPM onset were calculated. 11 related clinical characteristics, including age, race, marital status, grade, FIGO stage, radiotherapy, chemotherapy, and surgery, were further explore. Logistic and Cox regressions were employed to predict risk factors for CCSPMs diagnosis. Finally, two nomograms were developed to predict the probability occurrence and prognosis of CCSPMs, respectively. Results For diagnostic nomogram construction, 59,178 CC patients were randomly divided into training (n = 41,426) and validation cohorts (n = 17,752). For prognostic nomogram construction, 3,527 CCSPMs patients were randomly divided into training (n = 2,469) and validation cohorts (n = 1,058). The diagnostic nomogram consisting of above eleven independent risk factors (all P < 0.05), had high accuracy (AUCtraining = 0.851 and AUCvalidating = 0.845). The prognostic nomogram integrated with eight independent prognostic factors such as treatments, FIGO stage and TNM stage, performed well in predicting 5-year OS (AUCtraining = 0.835 and AUCvalidating = 0.837). Conclusion Our diagnostic and prognostic nomograms could facilitate clinicians to quantify individual SPMs risk and survival probabilities and optimize the surveillance recommendations and personalized clinical decision-making.