Frontiers in Oncology (Jul 2023)
Authentication of a survival nomogram for non-invasive micropapillary breast cancer
Abstract
PurposeWe aimed at establishing a nomogram to accurately predict the overall survival (OS) of non-metastatic invasive micropapillary breast carcinoma (IMPC).MethodsIn the training cohort, data from 429 patients with non-metastatic IMPC were obtained through the Surveillance, Epidemiology, and End Results (SEER) database. Other 102 patients were enrolled at the Xijing Hospital as validation cohort. Independent risk factors affecting OS were ascertained using univariate and multivariate Cox regression. A nomogram was established to predict OS at 3, 5 and 8 years. The concordance index (C-index), the area under a receiver operating characteristic (ROC) curve and calibration curves were utilized to assess calibration, discrimination and predictive accuracy. Finally, the nomogram was utilized to stratify the risk. The OS between groups was compared through Kaplan-Meier survival curves.ResultsThe multivariate analyses revealed that race (p = 0.047), surgery (p = 0.003), positive lymph nodes (p = 0.027), T stage (p = 0.045) and estrogen receptors (p = 0.019) were independent prognostic risk factors. The C-index was 0.766 (95% CI, 0.682-0.850) in the training cohort and 0.694 (95% CI, 0.527-0.861) in the validation cohort. Furthermore, the predicted OS was consistent with actual observation. The AUCs for OS at 3, 5 and 8 years were 0.786 (95% CI: 0.656-0.916), 0.791 (95% CI: 0.669-0.912), and 0.774 (95% CI: 0.688-0.860) in the training cohort, respectively. The area under the curves (AUCs) for OS at 3, 5 and 8 years were 0.653 (95% CI: 0.498-0.808), 0.683 (95% CI: 0.546-0.820), and 0.716 (95% CI: 0.595-0.836) in the validation cohort, respectively. The Kaplan-Meier survival curves revealed a significant different OS between groups in both cohorts (p<0.001).ConclusionOur novel prognostic nomogram for non-metastatic IMPC patients achieved a good level of accuracy in both cohorts and could be used to optimize the treatment based on the individual risk factors.
Keywords