AUA 2023: Development of a Longitudinal Prostate Cancer Transcriptomic and Real-World Clinical Data Linkage

( The 2023 AUA annual meeting included a session on markers in prostate cancer, featuring a presentation by Dr. Michael Leapman discussing the development of a longitudinal prostate cancer transcriptomic and real-world clinical data linkage. Genomic classifiers have been shown to be prognostic and impact clinical decision-making in retrospective studies. However, less is known about prognostic performance in the real-world clinical setting. Dr. Leapman and colleagues developed a novel linkage between the Decipher prostate genomic classifier and real-world patient data in the United States across payors and sites of care.

Clinical and transcriptomic data from clinical use of the Decipher prostate genomic classifier between 2013-2022 (Veracyte Inc., San Diego, CA) were linked with real-world data aggregated from insurance claims, pharmacy records, and Clarivate electronic health record data. Patients were anonymously linked between datasets by deterministic methods through a de-identification engine using encrypted tokens. The following is a schematic representation of data sources and the linkage structure: 

Flow chart clarivate datavant.jpg

The objective of this study was to develop algorithms for identifying prostate cancer diagnoses, treatment timing, and clinical outcomes (biochemical recurrence and prostate cancer metastases) in real-world data using diagnosis, CPT codes, pharmacy codes, SNOMED clinical terms, and unstructured text in the electronic health record. They then compared the accuracy of real-world data algorithms using clinical information obtained during Decipher testing as the reference standard.

 A total of 92,976 of 95,578 (97.2%) patients with Decipher prostate genomic classifier were successfully linked to real-world data, including 53,871 from biopsy and 39,105 from radical prostatectomy tests. The median age at Decipher testing was 66.4 years [IQR 61.0, 71.0]. The concordance of prostate cancer diagnoses was 85.0%, including 80.8% for biopsy and 90.7% for radical prostatectomy. Year of treatment was concordant in 98.5% of patients undergoing genomic classifier testing at radical prostatectomy, and 87.8% in patients with biopsy genomic classifier tests: 

RP Claim graph.jpg

Biochemical recurrence was identified based on diagnosis code (R 97.21 of ICD-10) (96.3%), unstructured text (0.05%), and both in (3.65%). Similarly, metastases were identified based on diagnosis codes (94.9%), unstructured text (1.38%), and both (3.73%). PSA verification was achieved through numerical laboratory data, and declines correlated with timing of radical prostatectomy and systemic therapy:


Dr. Leapman concluded his presentation by discussing the development of a longitudinal prostate cancer transcriptomic and real-world clinical data linkage with the following take-home messages:

  • This study establishes the first national-scale linkage of transcriptomic and longitudinal clinical data
  • Achieved high accuracy for identifying key clinical junctures including diagnosis, treatment, and early cancer outcome
  • This resource can be leveraged to enhance understanding of disease biology, patterns of care, and treatment effectiveness

Presented by: Michael Leapman, MD, MHS, Yale University, New Haven, CT

Written by: Zachary Klaassen, MD, MSc – Urologic Oncologist, Assistant Professor of Urology, Georgia Cancer Center, Augusta University/Medical College of Georgia, @zklaassen_md on Twitter during the 2023 American Urological Association (AUA) Annual Meeting, Chicago, IL, April 27 – May 1, 2023