Population-Based Prognostic Models for Head and Neck Cancers Using National Cancer Registry Data from Taiwan.

PURPOSE: This study aims to raise awareness of the disparities in survival predictions among races in head and neck cancer (HNC) patients by developing and validating population-based prognostic models specifically tailored for Taiwanese and Asian populations. METHODS: A total of 49,137 patients diagnosed with HNCs were included from the Taiwan Cancer Registry (TCR). Six prognostic models, divided into three categories based on surgical status, were developed to predict both overall survival (OS) and cancer-specific survival using the registered demographic and clinicopathological characteristics in the Cox proportional hazards model. The prognostic models underwent internal evaluation through a tenfold cross-validation among the TCR Taiwanese datasets and external validation across three primary racial populations using the Surveillance, Epidemiology, and End Results database. Predictive performance was assessed using discrimination analysis employing Harrell's c-index and calibration analysis with proportion tests. RESULTS: The TCR training and testing datasets demonstrated stable and favorable predictive performance, with all Harrell's c-index values ≥ 0.7 and almost all differences in proportion between the predicted and observed mortality being < 5%. In external validation, Asians exhibited the best performance compared with white and black populations, particularly in predicting OS, with all Harrell's c-index values > 0.7. CONCLUSIONS: Survival predictive disparities exist among different racial groups in HNCs. We have developed population-based prognostic models for Asians that can enhance clinical practice and treatment plans.

Distinct Survival Outcomes in Subgroups of Stage III Pancreatic Cancer Patients: Taiwan Cancer Registry and Surveillance, Epidemiology and End Results registry.

PURPOSE: Pancreatic cancer is one of the most malignant cancers with poor survival. The latest edition of the American Joint Committee on Cancer (AJCC) staging system classifies the majority of operable pancreatic cancer patients as stage-III, while dramatic heterogeneity is observed among these patients. Therefore, subgrouping is required to accurately predict their prognosis and define a treatment plan. This study conducts a cohort study to provide a more precise classification system for stage-III pancreatic cancer patients by utilizing clinical variables. METHODS: We analyzed survival using log-rank tests, univariate Cox-regression models, and Kaplan-Meier survival curves for stage-III pancreatic ductal adenocarcinoma (PDAC) patients from the Taiwan Cancer Registry (TCR). Patients were further divided into subgroups using classification and regression tree (CART) algorithm. All results were validated using the SEER database. RESULTS: Among stage-III PDAC patients, lymph node and tumor grade showed significant association with survival. Patients with N2 stage had higher mortality risks (hazard ratio [HR] = 2.30, 95% confidence interval [CI] 1.71-3.08, p < 0.0001) than N0 patients. Patients with grade 3 also had higher risk of mortality (HR = 3.80, 95% CI 2.25-6.39, p < 0.0001) than grade 1 patients. The CART algorithm stratified stage-III patients into four subgroups with significantly different survival rates. The median survival of the four subgroups was 23.5, 18.4, 14.5, and 9.0 months, respectively (p < 0.0001). Similar results were observed with SEER data. CONCLUSIONS: Lymph node involvement and tumor grade are predictive factors for survival in stage-III PDAC patients. This new precise classification system can be used to guide treatment planning in advanced-stage pancreatic cancer.

Predicting Colon Cancer-Specific Survival for the Asian Population Using National Cancer Registry Data from Taiwan.

PURPOSE: Colon cancer is the third most incident and life-threatening cancer in Taiwan. A comprehensive survival prediction system would greatly benefit clinical practice in this area. This study was designed to develop an accurate prognostic model for colon cancer patients by using clinicopathological variables obtained from the Taiwan Cancer Registry database. METHODS: We analyzed 20,218 colon cancer patients from the Taiwan Cancer Registry database, who were diagnosed between 2007 and 2015, were followed up until December 31, 2017, and had undergone curative surgery. We proposed two prognostic models, with different combinations of predictors. The first model used only traditional clinical features. The second model included several colon cancer site-specific factors (circumferential resection margin, perineural invasion, obstruction, and perforation), in addition to the traditional features. Both prediction models were developed by using a Cox proportional hazards model. Furthermore, we investigated whether race is a significant predictor of survival in colon cancer patients by using Model 1 on the Surveillance, Epidemiology, and End Results (SEER) cancer registry dataset. RESULTS: The proposed models displayed a robust prediction performance (all Harrell's c-index >0.8). For both the calibration and validation steps, the differences between the predicted and observed mortality were mostly less than 5%. CONCLUSIONS: The prediction model (Model 1) is an effective predictor of survival regardless of the ethnic background of patients and can potentially help to provide better prediction of colon cancer-specific survival outcomes, thus allowing physicians to improve treatment plans.

Risk evaluation of early-stage hormone receptor-positive and human epidermal growth factor receptor 2-negative breast cancer patients: a population-based study from Taiwan.

PURPOSE: To assess the prognostic value of the Dutch criteria for patients with early-stage hormone receptor-positive and human epidermal growth factor receptor 2-negative breast cancer from the Taiwan Cancer Registry Database. PATIENTS AND METHODS: We included 8295 patients with early-stage node-negative breast cancer who underwent surgery during January 2008-December 2012. Patients were stratified into low- and high-risk groups based on the Dutch criteria. The Kaplan-Meier method and log-rank test were used to estimate the difference in breast cancer-specific survival (BCSS) and overall survival (OS) between groups. Multivariable analysis was used to evaluate the prognostic value of the Dutch criteria. RESULTS: Overall, the low-risk and high-risk groups comprised 5375 and 2920 patients, respectively. In the low- and high-risk groups, the 5-year BCSS rate was 99.6% and 98.2% (P < 0.0001) and the 5-year OS rate was 98.3% and 96.8% (P < 0.0001), respectively. The hazard ratio for BCSS was 4.18 (95% confidence interval [CI] 2.63-6.63, P < 0.0001), and the hazard ratio for OS was 1.94 (95% CI 1.48-2.55); both were significantly poorer in the high-risk group than in the low-risk group. In the low-risk group, the 5-year BCSS and OS of patients who did and did not receive adjuvant chemotherapy were similar (99.5% versus 99.6% [P = 0.927] and 98.8% and 98.1% [P = 0.0683], respectively). CONCLUSION: The prognosis of low-risk patients as classified using the Dutch criteria is excellent with or without adjuvant chemotherapy. The benefit of multi-gene testing for chemotherapy decision-making might be minimal in these patients.

Development of a Gene-Based Prediction Model for Recurrence of Colorectal Cancer Using an Ensemble Learning Algorithm.

It is difficult to determine which patients with stage I and II colorectal cancer are at high risk of recurrence, qualifying them to undergo adjuvant chemotherapy. In this study, we aimed to determine a gene signature using gene expression data that could successfully identify high risk of recurrence among stage I and II colorectal cancer patients. First, a synthetic minority oversampling technique was used to address the problem of imbalanced data due to rare recurrence events. We then applied a sequential workflow of three methods (significance analysis of microarrays, logistic regression, and recursive feature elimination) to identify genes differentially expressed between patients with and without recurrence. To stabilize the prediction algorithm, we repeated the above processes on 10 subsets by bagging the training data set and then used support vector machine methods to construct the prediction models. The final predictions were determined by majority voting. The 10 models, using 51 differentially expressed genes, successfully predicted a high risk of recurrence within 3 years in the training data set, with a sensitivity of 91.18%. For the validation data sets, the sensitivity of the prediction with samples from two other countries was 80.00% and 91.67%. These prediction models can potentially function as a tool to decide if adjuvant chemotherapy should be administered after surgery for patients with stage I and II colorectal cancer.