Development and validation of a predictive nomogram for the specific mortality risk of luminal B breast cancer patients: a competing risk model based on real populations.

BACKGROUND

For clinical workers, disease-specific death is a better indicator of tumor severity. Breast cancer is the most prevalent malignancy in women. Luminol type B breast cancer is one of the biggest threats to women's health, and few studies have paid attention to its specific death. Early recognition of luminol type B breast cancer allows clinicians to assess the prognosis and develop more optimal treatment plans.

METHODS

In this study, the basic information of luminal B population, clinical and pathological characteristics, treatment regimen and survival data were collected from the SEER database. The patients were randomly divided into a training group and a validation group. The single-factor and multi-factor competitive risk models were used to analyze the independent influencing factors of tumor-specific death, and the predictive nomogram based on the competitive risk model was constructed. The consistency index (C-index) and calibration curves over time were used to evaluate the accuracy of the predicted nomograms.

RESULTS

This study included a total of 30,419 luminal B patient. The median follow-up period was 60 (IQR: 44-81) months. Among the 4,705 deaths during the follow-up period, 2,863 patients died specifically, accounting for 60.85% of the deaths. The independent predictive factors of cancer-specific mortality were: married, primary site, grade, stage, the primary site of operation, radiotherapy, chemotherapy, metastasis (lymph node, bone, brain, liver, lung), and Estrogen Receptor and Progesterone Receptor status. In the training cohort, the C-index of the predictive nomogram was 0.858, and the area under the receiver operating characteristic curve (AUC) for the first, third, and fifth years was 0.891, 0.864, and 0.845. The C-index of the validation cohort was 0.862, and the AUC for the first, third, and fifth years was 0.888, 0.872, and 0.849. The calibration curves of the training and validation cohorts showed that the predicted probability of the model was very consistent with the actual probability. And the 5-year survival rate according to the traditional survival analysis was 9.49%, while the 5-year specific mortality rate was only 8.88%.

CONCLUSIONS

The luminal B competing risk model we established has ideal accuracy and calibration.