Related Experiment Videos
Web-based application of integrated clinico-genomic prognostic nomograms in metastatic castration-resistant prostate
Muhammad Zaki Fadlullah1,2, Benjamin Haaland3, Liang Wang4
1Departments of Oncological Sciences and Biomedical Informatics, University of Utah-Huntsman Cancer Institute, Salt Lake City, UT, USA.
Background:
Survival in metastatic castration-resistant prostate cancer (mCRPC) is heterogeneous, and current prognostic models rely on clinical and non-tumor-specific laboratory variables. Plasma copy number alterations (CNAs) provide a blood-based measure of tumor biology for integrated clinico-genomic survival prediction in mCRPC.
Methods:
Plasma CNAs with prognostic significance were combined into a composite genomic prognostic score (CGPS) which was validated across three independent mCRPC cohorts (n = 561) for seven CNAs with gain/amplification in AR, MYC, and PIK3CA and loss/deletion in NKX3.1, TMPRSS2, TP53, ZBTB16. Multivariable Cox models with contemporary clinical prognostic factors alone, CGPS alone, and an integrated clinico-genomic model were built to estimate 24 and 36-month survivals. Model performances were assessed using the area under the curve (AUC) and freely accessible web-based applications were developed.
Results:
CGPS estimates across independent mCRPC cohorts showed high concordance with Pearson correlation coefficients of 0.86-0.97 for cross cohort comparisons. Time-dependent AUC values for the integrated clinico-genomic model at 24 and 36 months were 0.79 and 0.77, respectively, compared with 0.65 and 0.71 for using contemporary clinical factors alone and 0.74 and 0.69 for using CGPS alone. C-index was 0.73 for the integrated model compared with 0.70 for CGPS alone and 0.63 for clinical factors alone. The web application incorporates PSA, ALP, LDH, and CGPS and returns individualized survival probability estimates at 24 and 36 months.
Conclusion:
An integrated blood-based clinico-genomic prognostic model deployed as an open-access web application enables individualized survival estimation in mCRPC. This interactive tool could support prospective plasma CNA-based risk stratification in mCRPC.
