Introduction
With advancements in cancer screening, diagnosis, and treatment, the survival period of cancer patients has significantly increased. Long-term management of cancer patients should not only focus on the primary tumor but also monitor for the occurrence of secondary primary tumors (SPTs) and multiple primary tumors through regular check-ups and screenings. Secondary primary tumors are defined as new independent malignant tumors that arise in different locations following treatment for an initial malignancy, and they are not recurrences or metastases of the first primary tumor. Various factors may contribute to the development of SPTs, such as age, obesity, genetic susceptibility, the stage at diagnosis of the first primary tumor, and environmental exposure. Previous studies have shown that the incidence of SPTs in cancer patients ranges from 8-16%, varying by tumor site . SPTs pose a significant threat to cancer patients' survival, accounting for 55% of cancer-related deaths .Prostate cancer is the most common malignancy in men worldwide and is also a relatively common SPT. Approximately 10% of prostate cancer patients have a history of previous malignancy, and these patients have a worse prognosis compared to those without such a history [5,6]. However, due to the relatively low incidence of SPT prostate cancer (SPPCa), there is currently a lack of in-depth understanding of the clinical and prognostic characteristics of these patients, and high-quality clinical evidence for the treatment and management of SPPCa patients is scarce. Therefore, based on a large tumor population cohort from the US SEER (Surveillance, Epidemiology, and End Results) database, we conducted this study to analyze the clinical characteristics and prognostic factors of SPPCa patients, construct a prognostic prediction model, and develop a nomogram to provide clinical evidence for optimizing individualized treatment of SPPCa.
Methods
We screened SPPCa patients from the SEER database from 2010 to 2015, including a total of 5342 SPPCa cases. The median age of the patients was 68 years. The median follow-up time was 71 months. The most common primary tumor sites were the urogenital system (n=1527, 28.6%) and the digestive system (n=1283, 24.0%). Most SPPCa patients were in T1 stage (n=2708, 50.7%), N0 stage (n=5209, 97.5%), and M0 stage (n=5174, 96.7%). The majority of patients had a prostate-specific antigen (PSA) level between 4-10 ng/ml (n=3120, 58.4%) and a Gleason score in the ISUP1 group (n=2080, 38.9%). In terms of treatment, 1701 patients (31.8%) underwent surgery, 2054 patients (38.5%) received radiotherapy, and 33 patients (0.6%) underwent chemotherapy.
Results
Univariate and multivariate Cox analysis showed that age, the interval between the occurrence of multiple primary tumors, the site of the previous tumor, SPPCa tumor stage, PSA level, Gleason score, and treatment method were independent prognostic factors for overall survival (OS). For cancer-specific survival (CSS), age, tumor stage, PSA level, Gleason score, and SPPCa treatment method were independent prognostic factors. Kaplan-Meier analysis indicated that younger age, lower TNM stage, lower PSA level, lower Gleason score, and radical prostatectomy were associated with better survival outcomes. Patients with a history of respiratory system tumors had the shortest survival (5-year OS: 64.4%, 10-year OS: 38.3%; 5-year CSS: 91.7%, 10-year CSS: 84.0%), while those with a history of endocrine system tumors had the best prognosis (5-year OS: 91.6%, 10-year OS: 86.6%; 5-year CSS: 97.2%, 10-year CSS: 97.2%).
Model Construction
We divided the SPPCa patients into a training set and a validation set in a 7:3 ratio to construct and internally validate the model. We performed LASSO regression analysis (OS model: λ value 0.0027, CSS model: λ value 0.0013). Considering the results of multivariate Cox regression analysis and LASSO regression analysis, we included age, the interval between multiple primary tumors, tumor stage, PSA level, Gleason score, and treatment method in the OS prognostic prediction model. ROC curve analysis evaluated the OS prediction model, with AUC values for 3-year, 5-year, 7-year, and 9-year OS being 0.756, 0.744, 0.760, and 0.735, respectively. The C-index of the OS prediction model in the training set and validation set were 0.733 (95% CI: 0.724-0.741) and 0.722 (95% CI: 0.710-0.733), respectively. The CSS prognostic prediction model included age, PSA level, tumor stage, pathological grade, Gleason score, and treatment method. ROC curve analysis evaluated the CSS prediction model, with AUC values for 3-year, 5-year, 7-year, and 9-year CSS being 0.877, 0.854, 0.861, and 0.856, respectively. The C-index of the CSS prediction model in the training set and validation set were 0.838 (95% CI: 0.824-0.852) and 0.822 (95% CI: 0.800-0.844), respectively. According to the OS and CSS nomograms, the median risk score for OS in SPPCa patients was 91, and the median risk score for CSS was 79. Based on the OS and CSS risk scores, SPPCa patients were divided into high-risk and low-risk groups, and Kaplan-Meier analysis showed a significant difference in survival prognosis between the two groups. Calibration curves of the OS and CSS prediction models indicated good consistency between the nomogram predictions and actual observations.
Conclusion
This study analyzed the clinical characteristics of SPPCa based on a large-scale population cohort from the SEER database, identified prognostic factors for OS and CSS, and constructed prognostic prediction models. These models provide clinical insights for the prognostic assessment of SPPCa and offer clinical evidence for optimizing individualized treatment.
