ESOU 2022: State-of-the-art Lecture: Molecular Advances in Risk Stratification of Localized Prostate Cancer

(UroToday.com) The 2022 EAU Section of Oncological Urology (ESOU) Annual Meeting included a session on localized prostate cancer and a state-of-the-art lecture by Dr. Daniel Spratt discussing molecular advances in risk stratification of localized disease. Dr. Spratt notes that treatment decisions at their core are dominated by risk stratification, with the earliest risk stratification dating back to 1929 with the League of Nations’ World Health Organization developing the four-stage system for cervical cancer. Fast forward to the 1950s when Gleason grade grouping was added for prostate cancer, as well as the UICC publishing the first T, N, M staging classification. From the 1970s to the 1990s, the AJCC published editions 1 through 5, and the NCCN published their first prostate cancer guideline. Over the last two decades, we have seen the D’Amico risk stratification, AJCC 6^th, and 7^th edition, and the Zumsteg/Spratt (MSKCC) risk groups emerge, with many nomograms and multivariable models (ie. CAPRA) published along the way. More recently (2016 onwards), there has been the publication of the AJCC 8^th edition, the Spratt clinical genomic risk groups, and STAR-CAP for risk stratification.

However, Dr. Spratt cautions to not confuse motion with progress. Many clinicians still rely on a 60-year-old grading system (with modifications not specifically for prognostication), serum PSA (with its many limitations), and digital rectal examination (with prostate MRI only ~60% sensitive for T3 disease). Work from Dr. Spratt’s group presented at ASTRO 2021 looked at the prognostic and predictive performance of routine clinicopathologic variables enrolled on randomized phase III trials. The trials used for this individual patient data meta-analysis in terms of prognostic performance included RTOG 0126, RTOG 9910, RTOG 9408, RTOG 9413, RTOG 9202, EORTC 22991, EORTC 22863, EORTC 22961, Ottawa 0101, and DART 01/05 for a total of 10,535 patients. Using Gleason score, PSA, T-stage, and NCCN risk group, c-indices for biochemical recurrence and overall survival ranged from 0.52-0.60, slightly better than a coin-flip (0.50). Furthermore, looking at prognostic performance based on the Fine and Grey regression model and a random forest model, the c-indices for distant metastases were decent (~0.70) and modest at best for metastasis free survival and overall survival (both ~0.60):

Importantly, none of these models were developed/trained with the intent of optimal prognostication or prediction of treatment response. Dr. Spratt notes that in 2022 many scientists and clinicians still think of Gleason score as a clinical outcome. Thus, “you can’t learn anything new until you are open enough to forget everything you think you know.”

There is an objective versus subjective characterization of cancer: the cancer genotype being objective versus the cancer phenotype being subjective and error-prone, as highlighted in the following figure:

As such, there is a need to capture the vast molecular heterogeneity of prostate cancer, however not all gene expression tests appear to be equal. Dr. Spratt notes that for Oncotype, there is only one prospective trial in the active surveillance population, which was borderline significant on multivariable analysis for predicting adverse pathology.¹ For Prolaris, there are no prospective trials in radically treated patients, but there is a recent large retrospective study showing a c-index for CCP of only 0.52.² With regards to Decipher, this has been validated in multiple randomized clinical trials and prospective registries to prognosticate biochemical recurrence, metastasis free survival, and overall survival.³ In this systematic review of the evidence for the Decipher genomic classifier, Dr. Spratt notes that at the time of publication, there were 42 studies comprising more than 30,000 patients, in several disease states (localized prostate cancer, after radical prostatectomy, nmCRPC, and mHSPC). In addition to the endpoint of metastasis being evaluated (n=26 studies), other endpoints were also assessed, including adverse pathology (n=6), biochemical recurrence after radical prostatectomy (n=3), biochemical recurrence after radiotherapy (n=4), prostate cancer specific mortality (n=5), overall survival (n=3), and change in management (n=5).

Dr. Spratt then discussed the creation of the clinical-genomic risk groups. In his group’s seminal work published in 2018,⁴ the aim of their study was to develop a novel clinical-genomic risk grouping system that can readily be incorporated into treatment guidelines for localized prostate cancer. Over a median follow-up of 8 years for patients in the training cohort, 10-year distant metastasis rates for NCCN low, favorable-intermediate, unfavorable-intermediate, and high-risk were 7.3%, 9.2%, 38.0%, and 39.5%, respectively. In contrast, the three-tier clinical-genomic risk groups had 10-year distant metastasis rates of 3.5%, 29.4%, and 54.6%, for low-, intermediate-, and high-risk, respectively, which were consistent in the validation cohort. C-indices for the clinical-genomic risk grouping system (0.84; 95% CI, 0.61 to 0.93) were improved over NCCN (0.73; 95% CI, 0.60 to 0.86) and CAPRA (0.74; 95% CI, 0.65 to 0.84):

As such, a new “NCCN clinical genomic model” reclassifies 67% of patients:

44% of favorable-intermediate to low risk
41% of unfavorable intermediate to high risk
63% of high risk to very high risk

Furthermore, results for the Decipher genomic classifier have been analyzed in numerous randomized trials, in the localized, biochemically recurrent, and advanced disease states, with the most recent work from Dr. Spratt’s group scheduled to be presented at GU ASCO 2022. Published in 2021 in JAMA Oncology,⁵ Feng et al validated the Decipher genomic classifier in patients with recurrent prostate cancer using data from the NRG/RTOG 9601 randomized clinical trial. On multivariable analysis, the genomic classifier was independently associated with distant metastasis (HR 1.17, 95% CI 1.05-1.32; p = 0.006), prostate cancer specific mortality (HR 1.39, 95% CI, 1.20-1.63; p < 0.001), and overall survival (HR 1.17, 95% CI 1.06-1.29; p = 0.002) after adjusting for age, race/ethnicity, Gleason score, T stage, margin status, entry PSA, and treatment arm.

This extensive work using the Decipher genomic classifier has led to numerous phase III biomarker trials, including NRG GU1934 (GUIDANCE) and NRG GU009 (PREDICT-RT):

Dr. Spratt suggests that over the next several years, artificial intelligence for histology and radiology will continue changing the landscape of the objective vs subjective characterization of cancer.

Dr. Spratt concluded his presentation of molecular advances in risk stratification of localized prostate cancer with the following conclusions:

Nearly all treatment decisions in prostate cancer are based on prognosis
Current standard of care risk stratification tools have unacceptable performance, which inherently over- and under treat the vast majority of patients. We have improved models (ie. STAR-CAP), but these are still rarely used
Gene expression tests have proven to be superior in prognostication, however not all have equal data to support clinical use, and there are challenges in global adoption
A digital pathology and MRI-based approach may provide more cost-effective solutions

Presented by: Daniel E. Spratt, MD, Chair, Department of Radiation Oncology, Vincent K. Smith Professor, Case Western University, Cleveland, OH

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 2022 EAU Section of Oncological Urology (ESOU) Hybrid Annual Meeting, Madrid, Spain, Fri, Jan 21 – Sun, Jan 23, 2022.

References:

Login