- The decision to draw a PSA? – SNPs?
- 1st biopsy? – PCA3, phi, 4K, SelectMDx, ExoDx
- 2nd biopsy? – ConfirmMDx, PCA3, phi, 4K
- Pre-treatment? – OncoType, Prolaris, Decipher, ProMark
- Post-op treatment? – Decipher, Prolaris, OncoType
- Advanced disease – ARv7?
Specific to tests to consider before a first biopsy, 4K and phi were compared in 531 men from Sweden [1]. The four-kallikrein panel showed AUCs of 0.69 when predicting any-grade prostate cancer and 0.72 when predicting high-grade cancer (Gleason score ≥7). Similar values were found for PHI, corresponding to an AUC of 0.70 for any-grade disease and 0.71 for high-grade prostate cancer. Both models had higher AUCs than a base model with PSA value and age (p<0.0001 for both). SelectMDx is a urinary test for HOXC6 and DLX1 mRNA transcripts, which has been validated in two multi-center cohorts across 6 centers in the Netherlands, yielding a NPV of >95% for Gleason ≥7 prostate cancer. MiPS (PCA3 + TMPRSS2:ERG) can also yield NPVs of >95% for Gleason ≥7 prostate cancer, taking a PSA-only based model AUC from 0.64 to 0.75.
In the post-diagnosis setting, Cooperberg notes that we already have free, reliable predictive nomograms, as well as CAPRA/CAPRA-S. To improve on CAPRA, the bar is set high: CAPRA has a c-index of 0.79, even without taking into account PSA density, PSA kinetics, and pathology details (cribriform, ductal invasion, etc). Prognostic tests for newly diagnosed “low risk” patients include Prolaris (cell cycle progression score), OncoType (Genomic Prostate Score), and Decipher (genomic classifier). These are all based on RNA expression of gene sets derived from FPE biopsy tissue. Dr. Cooperberg notes that all of the above tests have shown and have been validated to improve multivariable model performance for post-treatment endpoints (adverse path, BCR, metastasis, CSM). With regards to urine tests, Dr. Cooperberg states that PCA3 and T2-ERG are associated with higher risk disease but not independently. Importantly, all “decision change” studies are extremely prone to bias and use the same historical controls and conflate an association with causality. Ultimately, future endeavors with better imaging, biomarkers and genomics can hopefully improve tailoring men for active surveillance – can we “undiagnose’ a subset of men. Cooperberg cautions that “prostate cancer risk cannot be dichotomized – it is not a pregnancy test that turns blue and we take out the prostate.”
In conclusion, Cooperberg highlights several take-home messages:
- Existing markers have independent prognostic value
- Pre-diagnostic markers may be excellent reflex tests (and likely outperform MRI)
- How to use post diagnostic tests in clinical practice (and how to integrate with novel imaging) is not yet clear
- Genomics will help us tailor active surveillance, perhaps not today but very soon
- We are only at the “end of the beginning”
1. Nordstrom T, Vickers A, Assel M, et al. Comparison between the four-kallikrein panel and prostate health index predicting prostate cancer. Eur Urol 2015;68(1):139-146.
Presented by: Matthew Cooperberg, MD, MPH, FACS, UCSF Helen Diller Family Comprehensive Cancer Center
Read the Opposing Debate: Biomarker Debate: Adopt, Expand, Trust (CON)
Written by: Zachary Klaassen, MD, Urologic Oncology Fellow, University of Toronto, Princess Margaret Cancer Centre at the 2018 AUA Annual Meeting - May 18 - 21, 2018 – San Francisco, CA USA