Predictors of Distant Metastasis and Prognosis in Newly Diagnosed T1 Intrahepatic Cholangiocarcinoma

Abstract

Background: According to American Joint Committee on Cancer (AJCC) 8th staging system, T1 intrahepatic cholangiocarcinoma (T1 ICC) is considered a tumor with no vascular invasion. However, T1 ICC usually occurs distant metastasis (DM), and the clinical features of these patients could help clinicians identify the high-risk population.

Methods: We reviewed 1959 newly diagnosed patients with T1 ICC from the Surveillance, Epidemiology, and End Results (SEER) database during 2004-2018. Logistic regression models and Cox proportional hazards models were conducted to predict the risk of DM and overall survival (OS), respectively, and then, web-based nomograms were constructed. Decision curve analysis (DCA) and clinical impact curves (CIC) were used to measure the clinical utility of the models. The low-, medium-, and high-risk groups were identified by calculating the summary of the risk points. Nomograms on the web were also created to help clinicians better use these prediction models.

Results: Tumor size and lymph node metastasis accounted for the first two largest proportions among the DM nomogram scores, while surgery, DM, age at diagnosis, chemotherapy, and lymph node metastasis occupied the largest percentage in OS nomogram. DM nomogram was established for these newly diagnosed patients with T1 ICC, and OS nomogram was developed to visually predict the OS rate of 3, 5, and 10 years. The calibration curves revealed a valid predictive accuracy of nomograms, of which the C-index was 0.703 and 0.740, respectively, for good discrimination. DCAs, CICs, and risk subgroups showed the clinical validity of these nomograms. Two websites were created to make it easier to use these nomograms.

Conclusions: Novel web-based nomograms predicting the risk of DM and OS for T1 ICC were constructed. These predictive tools might help clinicians make precise clinical strategies for each patient with T1 ICC.

Figure 1

Analytical cohort and exclusion criteria of T1 ICC patients.

Figure 2

Nomogram, calibration curve, decision curve analysis, and clinical impact curve for predicting distant metastasis (DM) in patients with T1 ICC. There are four factors in DM prediction nomogram (a). Calibration curve (b) for predicting DM is shown, and C-index was 0.703 in the training cohort and 0.716 in the validation cohort. The decision curve (c) of the nomogram predicting DM was plotted. The x-axis represents the threshold probability and the y-axis represents the net benefit. The horizontal blue line represents one extreme situation that no patients suffered DM, and the black line represents that all patients experience DM. Clinical impact curve (d) shows that the number of high-risk patients and the number of high-risk patients with event were plotted by different threshold probability in a population.

Figure 3

Nomogram and calibration curve for predicting overall survival (OS) in patients with T1 ICC. There are nine factors in OS prediction nomogram (a). Calibration curve for predicting 3-, 5-, and 10-year OS in the training cohort (b). Calibration curve for predicting 3-, 5-, and 10-year OS in the validation cohort (c).

Figure 4

Clinical effects of the risk score in the nomogram. Based on the quartile of risk score, nomograms divided patients into low-, middle-, and high-risk subgroups, respectively. Clinical utility of these subgroups for predicting DM is presented by constituent ratio (a). The Kaplan-Meier method is used to find out the significance among these risk subgroups (b).

Figure 5

The decision curve of the nomogram predicting 3-, 5-, and 10-year OS in the training (a) and validation (b) cohorts was plotted. The x-axis represents the threshold probability, and the y-axis represents the standardized net benefit. The horizontal black line represents one extreme situation that all patients were alive, and the grey line represents the other extreme situation that all patients were dead.