Biomarkers have a lot of promise for new tumor detection, molecular staging, detecting recurrence and selecting appropriate treatment. There are a variety of biomarker sources, including biofluids (urine, blood) and tissue. As a result, there are numerous biomarkers reported, but the level of evidence is not robust enough for most of them.
Prior to delving into specific biomarkers, Dr. Inman reviewed some key concepts that make the identification of clinically relevant biomarkers difficult.
1) Biomarker development goes through 6 phases (preclinical, phases 0-4). Most biomarkers have made it through phase 1-2, very rarely have any of them made it to phase III.
2) Pretest probability is very important in determining utility of a biomarker – and many studies don’t account for it.
- Pretest probability limits the utility of any biomarker
- Tests with excellent accuracy still suffer
3) Predictive vs. prognostic ability
- Predictive indicates that the biomarker positivity is associated with a clinical response
- Prognostic biomarkers provides information about the patient's overall cancer outcome, regardless of therapy
- Biomarkers can be both predictive and prognostic
4) What is the gold standard? Cystoscopy or TURBT?
5) What to do with equivocal results?
6) Spectrum bias
- Tests perform differently in different patient cohorts
- Therefore, need to be clear about which patient population you are testing
7) Anticipatory positives
- If the gold standard is negative, and biomarker is positive, it may predict future development of disease
- However, some portion of patients with negative biomarker may also develop BCa
- Need to account for that in the predictive value – but most authors don’t
After reviewing the EAU, NICE, AUA/SUO and NCCN guidelines for the following scenarios (hematuria evaluation, NMIBC surveillance, Molecular staging, BCG evaluation, chemotherapy response and immunotherapy response), he summarized the groups guidelines regarding biomarkers with the following slide:
He did provide specific examples along the way, a few of which I will highlight below:
1) NMIBC surveillance
- numerous markers evaluated with goal of replacing WLC cystoscopy
- However, none (except for CyVision in New Zealand) have succeeded
- They are all stage and grade dependent
- They all had similar sensitivity and specificity – specificity was always higher
2) Biomarkers often look promising in early stages (retrospective series)
- p53 as a molecular marker was a good example
- In early retrospective series, it was a strong predictor of relapse-free survival after cystectomy
- However, during a prospective trial, it was neither prognostic nor predictive!
- Trial validation is crucial!
3) Decipher predicts response to chemotherapy
- Recent study looking at the different molecular subtypes and their Decipher scores
- While 3 of the 4 subtypes had no change in response to chemotherapy, patients with basal type BCa had strong response to neoadjuvant chemotherapy
4) PDL-1 is the caution tale about setting up a trial eligibility based on a biomarker
- IMvigor211 – based on prior atezo and nivolumab trials, setup to determine efficacy of PDL-1 as a biomarker of response
- While on overall ITT analysis, there was a survival benefit to Atezo over chemotherapy, in the patients with strong PDL-1 expression, there was no difference
- Killed its utility as a biomarker
Presented by: Brant Inman
Written by: Thenappan Chandrasekar, MD, Clinical Fellow, University of Toronto, Twitter: @tchandra_uromd at the 37th Congress of Société Internationale d’Urologie - October 19-22, 2017- Lisbon, Portugal