Development of a Longitudinal Prostate Cancer Transcriptomic and Clinical Data Linkage

Area of Science:

• Oncology
• Genomics
• Health Informatics

Background:

• Tissue-based gene expression testing is crucial for prostate cancer risk stratification.
• The prognostic performance of genomic classifiers (GC) in real-world clinical settings requires further understanding.

Purpose of the Study:

• To establish a national-scale linkage between prostate cancer genomic classifier (GC) data and diverse clinical data sources across the US.
• To assess the accuracy of algorithms for identifying key clinical events and outcomes using linked data.

Main Methods:

• A cohort study linked transcriptomic data from prostate GC testing (2016-2022) with insurance claims, pharmacy records, and EHR data.
• Deterministic linkage methods and refined algorithms were used to identify prostate cancer diagnoses, treatment timing, and outcomes.
• Sensitivity analyses were performed using clinical and pathological GC testing data as the reference standard.

Main Results:

• Over 97% of participants (92,976/95,578) were successfully linked.
• High sensitivity was achieved for identifying prostate cancer diagnoses (85.0%) and treatment timing concordance (97.9% for radical prostatectomy).
• The linkage demonstrated the feasibility of integrating genomic and clinical data for large-scale analysis.

Conclusions:

• A national-scale linkage of transcriptomic and longitudinal clinical data was successfully established with high accuracy.
• This integrated data resource can significantly enhance understanding of prostate cancer biology, care patterns, and treatment effectiveness.
• The findings support the use of linked data for real-world evidence generation in prostate cancer research.