Related Experiment Video
Updated: Jun 4, 2026
Establishing a Competing Risk Regression Nomogram Model for Survival Data
Published on: October 23, 2020
Clinico-genomic nomograms for estimation of survival risk in metastatic castrate-resistant prostate cancer
Manish Kohli1, Anushka Shankar1, Jennifer Lloyd2
1Division of Oncology, Department of Medicine, Huntsman Cancer Institute, University of Utah, Salt Lake City, UT, United States.
Background:
To enhance risk stratification in metastatic castrate-resistant prostate cancer (mCRPC), we developed clinical and integrated clinico-genomic prognostic nomograms combining clinical prognostic factors with copy number alteration-based risk scores (RSs) in circulating tumor DNA (ctDNA) and metastatic tissue to predict 1-, 2-, and 3-year overall survival (OS) probabilities.
Methods:
A clinical prognostic nomogram was developed from publicly accessed Yale University Open Data Access (YODA) data-sciences project (N = 1088 patients). A second publicly available independent dataset (N = 82) with an 11-gene prognostic RS developed concurrently obtained ctDNA and matched metastatic tissue specimens (gains: AR, MYC, COL22A1, PIK3CA, PIK3CB, NOTCH1; losses: TMPRSS2, NCOR1, ZBTB18, TP53, NKX3-1) was used to develop an integrated clinico-genomic survival nomogram. Independent predictors for survival were identified in each dataset using univariate and multivariate Cox proportional hazard. Multivariate regression coefficients with statistical significance (P < .05) were used to develop prognostic nomograms to estimate 1-, 2- and 3-year OS probabilities. Nomogram performances were assessed using time-dependent area under curves (t-AUCs). All analyses were conducted in RStudio (v4.1.2).
Results:
The YODA dataset t-AUCs for the prognostic nomogram for estimating OS at 1, 2, and 3 years were 0.74, 0.70, and 0.74. In the independent second dataset, the integrated clinico-genomic prognostic nomogram was observed to have higher t-AUCs for predicting 3-year mCRPC survival at 0.807 compared with 0.714 using clinical factors alone.
Conclusions:
An integrated clinico-genomic nomogram with ctDNA RSs achieved the highest t-AUC for 3-year survival in mCRPC and may enable precision in identification of poor-prognosis subgroups based on somatic alterations.
