The miR Scientific Sentinel® Prostate Cancer Classifier Platform (Sentinel® PCC4 Test) is a new molecular test that interrogates 442 small non-coding RNAs extracted from urinary exosomes. The Sentinel® PCC4 Test provides urologists with an early, non-invasive and precise indication of disease status in men presenting with clinical suspicion of prostate cancer. The test provides an initial classification of disease status as either no molecular evidence of prostate cancer (NMEPC), or molecular evidence (MEPC) of low-, intermediate- or high-risk of aggressive prostate cancer. The current standard of care (SOC) assesses the risk of prostate cancer based on information that includes a suspicious DRE, and "elevated" serum PSA (commonly greater than 3ng/mL based on the current NCCN guidelines, but in clinical practice can frequently be less than or greater than 3ng/mL, depending on the practice). The estimated sensitivity (44%) and specificity (68%) of the current SOC means that a significant portion of prostate cancer cases are not identified, and many biopsies with resulting co-morbidities are unnecessary.
Patients who enroll in this study will receive the current SOC and have a TRUS- or MRI-guided fusion core needle biopsy. Biopsies will be read by pathologists associated with each participating clinical institution at which the biopsy was performed. The Gleason score, tumor staging, and anatomic location for each positive core will be collected at each site, and barcoded urine samples will be couriered overnight to the miR Scientific laboratories for Sentinel® PCC4 Test analysis. To minimize the potential for inter-observer variability, the pathology of the core needle biopsies for each patient will also be reviewed by an independent central pathologist, not associated with miR Scientific.
After assessing the pathology, anatomic contribution and additional clinical data (particularly DRE and serum PSA levels) that inform Grade Group assessment and the NCCN risk assessment, the final risk group assessment for both Primary Study Objective #1 and Primary Study Objective #2 will be determined by the pathologist. The final risk group determination for each patient will be used individually to represent the risk status provided by the current SOC for comparison to the molecular classifications provided by the Sentinel® PCC4 Test.
The performance of the Sentinel® PCC4 Test to classify the disease status of a future patient (with unknown true disease status) as either no molecular evidence of prostate cancer or, for patients with molecular evidence of prostate cancer, as having low-risk, intermediate-risk or high-risk of aggressive prostate cancer will be established for both Primary Study Objectives.
Primary Study Objective #1: To establish the performance characteristics of the Sentinel® Prostate Cancer Classifier Platform (Sentinel® PCC4 Test) using pathological grading (Grade Grouping based on Gleason score as proposed by Epstein and colleagues in 2013) as the comparative basis for assessing the prostate health of men aged 45-75 years presenting with suspicion of prostate cancer.
The performance characteristics of the Sentinel® PCC4 Test as described via the Primary Study Objectives will be assessed in comparison to core needle biopsy report using pathology Gleason score grading (Grade Group) risk classifiers where GG1 is indicative or low-risk, GG2 as intermediate risk and GG3-5 as high-risk of aggressive prostate cancer. The sensitivity, specificity, positive and negative predictive values of the Sentinel® PCC4 Test for risk-classification in concordance with the GG-risk assessment will be established.
Primary Study Objective #2: To establish the performance characteristics of the Sentinel® PCC4 Test when using the modified risk assessment criteria recently recommended by the National Comprehensive Cancer Network (NCCN) as the basis for assessing the prostate health of men aged 45-75 years presenting with suspicion of prostate cancer.
In addition to the pathological grade of the tumor, the NCCN risk assessment incorporates additional clinical measures associated with tumor progression, including number of positive cores, percent tumor involvement per core, tumor stage (anatomical location of positive cores), serum PSA and PSA density. The same molecular expression data from the Sentine® PCC4 Test will be used to evaluate both Primary Objectives for each patient.
Statistical Methodology: This study will enroll patients visiting urology clinics for initial clinical workup based on any clinical suspicion of prostate cancer. The study will evaluate the properties of the Sentinel® PCC4 Test that is based on use of a proprietary classification system to identify patients with no molecular evidence of prostate cancer and to initially classify the risk of prostate cancer patients.
For a given two-group comparison, the proprietary Classification Algorithm controls sensitivity for a patient with unknown true disease status at or above a pre-specified level, denoted 1-α; for example, the value that has been assumed in this design is α=0.05. Note that the value of α represents the false-negative rate of the test for the given two-group comparison.
The initial nominal labeling of clinical groups will be done separately when using the Gleason Grade group (Primary Study Objective #1) and recent NCCN definitions (Primary Study Objective #2) as described in detail above. Cross-validation will be used in the training dataset to determine the classification cut-offs for a given two-group comparison in the Sentinel® PCC4 Test. A general description of this procedure is as follows. For each participant in the training dataset (current participant), a two-group classification for this current participant will be determined based on the remaining members of the training dataset using only the current participant's sncRNA sequence; the accuracy of this determination will be assessed by comparison of the group classification determined to the actual label given to the current participant (either Grade Group or NCCN group). That is, the disease status label of the current participant in the training dataset will be blinded, thereby mimicking the setting for classification of a future patient. This procedure is then repeated separately for each subject in the training dataset. The tuning parameters and cutoff boundary used for each of the two-group classifications comprising the Sentinel® PCC4 Test are then calculated so that the empirical sensitivity over participants in the training dataset is at least 1-α.
With these cut-offs and classification rules determined using the training dataset for the Sentinel® PCC4 Test, the Sentinel® PCC4 classifications for each member of the validation dataset will be determined using only information generated from their expression levels of the sncRNA sequences; the corresponding results will be stored in a date-stamped, locked file as described above. The clinical risk group for each member of the validation dataset as determined by the central pathologist will also be stored in a locked and date-stamped file. When both files are completed, they will be released to the miR Scientific team and the performance characteristics of the Sentinel® PCC4 Test established in the validation dataset; note that these error rates refer to the classification of a future patient with unknown disease status.
