Prostate Health Index and multiparametric magnetic resonance imaging to predict prostate cancer grade reclassification in active surveillance.

OBJECTIVE: To identify the value of combining the Prostate Health Index (PHI) and multiparametric magnetic resonance imaging (mpMRI), tools which have previously been shown to be independently predictive of prostate cancer (PCa) grade reclassification (GR; Gleason score >6), for the purpose of predicting GR at the next surveillance biopsy to reduce unnecessary prostate biopsies for men in PCa active surveillance (AS). PATIENTS AND METHODS: Between 2014 and 2019, we retrospectively identified 253 consecutive men in the Johns Hopkins AS programme who had mpMRI and PHI followed by a systematic ± targeted biopsy. PHI and PHI density (PHID) were evaluated across Prostate Imaging-Reporting and Data System version 2.0 (PI-RADSv2) scores and compared to those with and without GR. Next, the negative predictive value (NPV) and area under the receiver operating curve (AUC) were calculated to compare the diagnostic value of PI-RADSv2 score combined with PHI, PHID, or prostate-specific antigen density (PSAD) for GR using their respective first quartile as a cut-off. RESULTS: Of the 253 men, 38 men (15%) had GR. Men with GR had higher PHI values (40.7 vs 32.0, P = 0.001), PHID (0.83 vs 0.57, P = 0.007), and PSAD (0.12 vs 0.10, P = 0.037). A PI-RADSv2 ≤3 alone had a NPV of 91.6% for GR (AUC 0.67). Using a PHI cut-off of 25.6 in addition to PI-RADSv2 ≤3, the NPV and AUC were both increased to 98% and 0.70, respectively. Using a PSAD cut-off of 0.07 ng/mL/mL with PI-RADSv2 had an AUC of 0.69 and NPV of 95.4%. PHI and PI-RADSv2 together could have avoided 20% of biopsies at the cost of missing 2.6% of GRs. CONCLUSIONS: The combination of PHI and mpMRI can aid in the prediction of GR in men on AS and may be useful for decreasing the burden of surveillance prostate biopsies.

Germline mutations in PPFIBP2 are associated with lethal prostate cancer.

BACKGROUND: Few genes have germline mutations which predispose men to more aggressive prostate cancer (PCa). This study evaluated the contribution of germline loss of function (LOF) variants in PPFIBP2 to risk of lethal PCa. METHODS: A case-case study of 1414 PCa patients with lethal PCa and low-risk localized PCa was performed. Germline DNA samples from these patients were sequenced for PPFIBP2. Mutation carrier rates and association with lethal PCa were analyzed using the Fisher exact test, logistic regression, and Kaplan-Meier survival analysis. RESULTS: In the entire study population, eight patients, all of European ancestry, were identified as carrying PPFIBP2 pathogenic or likely pathogenic mutations. Seven (1.52%) of 462 lethal PCa patients were carriers compared with only one (0.12%) carrier in 810 low-risk PCa patients, P = 0.0029. The estimated Odds Ratio (OR) of carrying PPFIBP2 mutation for lethal PCa was 13.8 in European American population. The PPFIBP2 loss-of-function mutation carrier rate in lethal PCa cases was also higher than in 33 370 non-Finnish European individuals from the Exome Aggregation Consortium (ExAC) (carrier rate of 0.17%, P = 1.92 × 10-5 ) and in 498 men with localized PCa from The Cancer Genome Atlas cohort (TCGA) cohort (carrier rate of 0%, P = 0.0058). Survival analysis in European American lethal cases revealed PPFIBP2 mutation status as an independent predictor of shorter survival after adjusting for age at diagnosis, PSA at diagnosis, and genetic background (hazard ratio = 2.62, P = 0.034). CONCLUSIONS: While larger studies are needed, germline mutations in a novel gene, PPFIBP2, differentiated risk for lethal PCa from low-risk cases and were associated with shorter survival times after diagnosis.

Qualitative study on decision-making by prostate cancer physicians during active surveillance.

OBJECTIVE: To explore and identify factors that influence physicians' decisions while monitoring patients with prostate cancer on active surveillance (AS). SUBJECTS AND METHODS: A purposive sampling strategy was used to identify physicians treating prostate cancer from diverse clinical backgrounds and geographic areas across the USA. We conducted 24 in-depth interviews from July to December 2015, until thematic saturation was reached. The Applied Thematic Analysis framework was used to guide data collection and analysis. Interview transcripts were reviewed and coded independently by two researchers. Matrix analysis and NVivo software were used for organization and further analysis. RESULTS: Eight key themes emerged to explain variation in AS monitoring: (i) physician comfort with AS; (ii) protocol selection; (iii) beliefs about the utility and quality of testing; (iv) years of experience and exposure to AS during training; (v) concerns about inflicting 'harm'; (vi) patient characteristics; (vii) patient preferences; and (viii) financial incentives. CONCLUSION: These qualitative data reveal which factors influence physicians who manage patients on AS. There is tension between providing standardized care while also considering individual patients' needs and health status. Additional education on AS is needed during urology training and continuing medical education. Future research is needed to empirically understand whether any specific protocol is superior to tailored, individualized care.

A Bayesian hierarchical model for prediction of latent health states from multiple data sources with application to active surveillance of prostate cancer.

In this article, we present a Bayesian hierarchical model for predicting a latent health state from longitudinal clinical measurements. Model development is motivated by the need to integrate multiple sources of data to improve clinical decisions about whether to remove or irradiate a patient's prostate cancer. Existing modeling approaches are extended to accommodate measurement error in cancer state determinations based on biopsied tissue, clinical measurements possibly not missing at random, and informative partial observation of the true state. The proposed model enables estimation of whether an individual's underlying prostate cancer is aggressive, requiring surgery and/or radiation, or indolent, permitting continued surveillance. These individualized predictions can then be communicated to clinicians and patients to inform decision-making. We demonstrate the model with data from a cohort of low-risk prostate cancer patients at Johns Hopkins University and assess predictive accuracy among a subset for whom true cancer state is observed. Simulation studies confirm model performance and explore the impact of adjusting for informative missingness on true state predictions. R code is provided in an online supplement and at http://github.com/rycoley/prediction-prostate-surveillance.