A Scaled Proteomic Discovery Study for Prostate Cancer Diagnostic Markers Using ProteographTM and Trapped Ion Mobility Mass Spectrometry.

There is a significant unmet need for clinical reflex tests that increase the specificity of prostate-specific antigen blood testing, the longstanding but imperfect tool for prostate cancer diagnosis. Towards this endpoint, we present the results from a discovery study that identifies new prostate-specific antigen reflex markers in a large-scale patient serum cohort using differentiating technologies for deep proteomic interrogation. We detect known prostate cancer blood markers as well as novel candidates. Through bioinformatic pathway enrichment and network analysis, we reveal associations of differentially abundant proteins with cytoskeletal, metabolic, and ribosomal activities, all of which have been previously associated with prostate cancer progression. Additionally, optimized machine learning classifier analysis reveals proteomic signatures capable of detecting the disease prior to biopsy, performing on par with an accepted clinical risk calculator benchmark.

Targeted Mass Spectrometry Assays for Specific Quantification of Urinary proPSA Isoforms.

Prostate cancer (PCa) is the second leading cause of male cancer-related deaths in the United States. The pre-mature forms of prostate-specific antigen (PSA), proPSA, were shown to be associated with PCa. However, there is a technical challenge in the development of antibody-based immunoassays for specific recognition of each individual proPSA isoform. Herein, we report the development of highly specific, antibody-free, targeted mass spectrometry assays for simultaneous quantification of [-2], [-4], [-5], and [-7] proPSA isoforms in voided urine. The newly developed proPSA assays capitalize on Lys-C digestion to generate surrogate peptides with appropriate length (9-16 amino acids) along with long-gradient liquid chromatography separation. The assay utility of these isoform markers was evaluated in a cohort of 30 well-established clinical urine samples for distinguishing PCa patients from healthy controls. Under the 95% confidence interval, the combination of [-2] and [-4] proPSA isoforms yields the area under curve (AUC) of 0.86, and the AUC value for the combined all four isoforms was calculated to be 0.85. We have further verified [-2]proPSA, the dominant isoform, in an independent cohort of 34 clinical urine samples. Validation of proPSA isoforms in large-scale cohorts is needed to demonstrate their potential clinical utility.

Accommodating heterogeneous missing data patterns for prostate cancer risk prediction.

BACKGROUND: We compared six commonly used logistic regression methods for accommodating missing risk factor data from multiple heterogeneous cohorts, in which some cohorts do not collect some risk factors at all, and developed an online risk prediction tool that accommodates missing risk factors from the end-user. METHODS: Ten North American and European cohorts from the Prostate Biopsy Collaborative Group (PBCG) were used for fitting a risk prediction tool for clinically significant prostate cancer, defined as Gleason grade group ≥ 2 on standard TRUS prostate biopsy. One large European PBCG cohort was withheld for external validation, where calibration-in-the-large (CIL), calibration curves, and area-underneath-the-receiver-operating characteristic curve (AUC) were evaluated. Ten-fold leave-one-cohort-internal validation further validated the optimal missing data approach. RESULTS: Among 12,703 biopsies from 10 training cohorts, 3,597 (28%) had clinically significant prostate cancer, compared to 1,757 of 5,540 (32%) in the external validation cohort. In external validation, the available cases method that pooled individual patient data containing all risk factors input by an end-user had best CIL, under-predicting risks as percentages by 2.9% on average, and obtained an AUC of 75.7%. Imputation had the worst CIL (-13.3%). The available cases method was further validated as optimal in internal cross-validation and thus used for development of an online risk tool. For end-users of the risk tool, two risk factors were mandatory: serum prostate-specific antigen (PSA) and age, and ten were optional: digital rectal exam, prostate volume, prior negative biopsy, 5-alpha-reductase-inhibitor use, prior PSA screen, African ancestry, Hispanic ethnicity, first-degree prostate-, breast-, and second-degree prostate-cancer family history. CONCLUSION: Developers of clinical risk prediction tools should optimize use of available data and sources even in the presence of high amounts of missing data and offer options for users with missing risk factors.

Prediction of future risk of any and higher-grade prostate cancer based on the PLCO and SELECT trials.

BACKGROUND: A model was built that characterized effects of individual factors on five-year prostate cancer (PCa) risk in the Prostate, Lung, Colon, and Ovarian Cancer Screening Trial (PLCO) and the Selenium and Vitamin E Cancer Prevention Trial (SELECT). This model was validated in a third San Antonio Biomarkers of Risk (SABOR) screening cohort. METHODS: A prediction model for 1- to 5-year risk of developing PCa and Gleason > 7 PCa (HG PCa) was built on PLCO and SELECT using the Cox proportional hazards model adjusting for patient baseline characteristics. Random forests and neural networks were compared to Cox proportional hazard survival models, using the trial datasets for model building and the SABOR cohort for model evaluation. The most accurate prediction model is included in an online calculator. RESULTS: The respective rates of PCa were 8.9%, 7.2%, and 11.1% in PLCO (n = 31,495), SELECT (n = 35,507), and SABOR (n = 1790) over median follow-up of 11.7, 8.1 and 9.0 years. The Cox model showed higher prostate-specific antigen (PSA), BMI and age, and African American race to be associated with PCa and HGPCa. Five-year risk predictions from the combined SELECT and PLCO model effectively discriminated risk in the SABOR cohort with C-index 0.76 (95% CI [0.72, 0.79]) for PCa, and 0.74 (95% CI [0.65,0.83]) for HGPCa. CONCLUSIONS: A 1- to 5-year PCa risk prediction model developed from PLCO and SELECT was validated with SABOR and implemented online. This model can individualize and inform shared screening decisions.

African-specific improvement of a polygenic hazard score for age at diagnosis of prostate cancer.

Polygenic hazard score (PHS) models are associated with age at diagnosis of prostate cancer. Our model developed in Europeans (PHS46) showed reduced performance in men with African genetic ancestry. We used a cross-validated search to identify single nucleotide polymorphisms (SNPs) that might improve performance in this population. Anonymized genotypic data were obtained from the PRACTICAL consortium for 6253 men with African genetic ancestry. Ten iterations of a 10-fold cross-validation search were conducted to select SNPs that would be included in the final PHS46+African model. The coefficients of PHS46+African were estimated in a Cox proportional hazards framework using age at diagnosis as the dependent variable and PHS46, and selected SNPs as predictors. The performance of PHS46 and PHS46+African was compared using the same cross-validated approach. Three SNPs (rs76229939, rs74421890 and rs5013678) were selected for inclusion in PHS46+African. All three SNPs are located on chromosome 8q24. PHS46+African showed substantial improvements in all performance metrics measured, including a 75% increase in the relative hazard of those in the upper 20% compared to the bottom 20% (2.47-4.34) and a 20% reduction in the relative hazard of those in the bottom 20% compared to the middle 40% (0.65-0.53). In conclusion, we identified three SNPs that substantially improved the association of PHS46 with age at diagnosis of prostate cancer in men with African genetic ancestry to levels comparable to Europeans.

Comparison of rectal swab, glove tip, and participant-collected stool techniques for gut microbiome sampling.

BACKGROUND: Studies of the gut microbiome are becoming increasingly important. Such studies require stool collections that can be processed or frozen in a timely manner so as not to alter the microbial content. Due to the logistical difficulties of home-based stool collection, there has been a challenge in selecting the appropriate sample collection technique and comparing results from different microbiome studies. Thus, we compared stool collection and two alternative clinic-based fecal microbiome collection techniques, including a newer glove-based collection method. RESULTS: We prospectively enrolled 22 adult men from our prostate cancer screening cohort SABOR (San Antonio Biomarkers of Risk for prostate cancer) in San Antonio, TX, from 8/2018 to 4/2019. A rectal swab and glove tip sample were collected from each participant during a one-time visit to our clinics. A single stool sample was collected at the participant's home. DNA was isolated from the fecal material and 16 s rRNA sequencing of the V1-V2 and V3-V4 regions was performed. We found the gut microbiome to be similar in richness and evenness, noting no differences in alpha diversity among the collection methods. The stool collection method, which remains the gold-standard method for the gut microbiome, proved to have different community composition compared to swab and glove tip techniques (p< 0.001) as measured by Bray-Curtis and unifrac distances. There were no significant differences in between the swab and glove tip samples with regard to beta diversity (p> 0.05). Despite differences between home-based stool and office-based fecal collection methods, we noted that the distance metrics for the three methods cluster by participant indicating within-person similarities. Additionally, no taxa differed among the methods in a Linear Discriminant Analysis Effect Size (LEfSe) analysis comparing all-against-all sampling methods. CONCLUSION: The glove tip method provides similar gut microbiome results as rectal swab and stool microbiome collection techniques. The addition of a new office-based collection technique could help easy and practical implementation of gut microbiome research studies and clinical practice.

A genome-wide association study of prostate cancer in Latinos.

Latinos represent <1% of samples analyzed to date in genome-wide association studies of cancer. The clinical value of genetic information in guiding personalized medicine in populations of non-European ancestry will require additional discovery and risk locus characterization efforts across populations. In the present study, we performed a GWAS of prostate cancer (PrCa) in 2,820 Latino PrCa cases and 5,293 controls to search for novel PrCa risk loci and to examine the generalizability of known PrCa risk loci in Latino men. We also conducted a genetic admixture-mapping scan to identify PrCa risk alleles associated with local ancestry. Genome-wide significant associations were observed with 84 variants all located at the known PrCa risk regions at 8q24 (128.484-128.548) and 10q11.22 (MSMB gene). In admixture mapping, we observed genome-wide significant associations with local African ancestry at 8q24. Of the 162 established PrCa risk variants that are common in Latino men, 135 (83.3%) had effects that were directionally consistent as previously reported, among which 55 (34.0%) were statistically significant with p < 0.05. A polygenic risk model of the known PrCa risk variants showed that, compared to men with average risk (25th-75th percentile of the polygenic risk score distribution), men in the top 10% had a 3.19-fold (95% CI: 2.65, 3.84) increased PrCa risk. In conclusion, we found that the known PrCa risk variants can effectively stratify PrCa risk in Latino men. Larger studies in Latino populations will be required to discover and characterize genetic risk variants for PrCa and improve risk stratification for this population.

Multi-cohort modeling strategies for scalable globally accessible prostate cancer risk tools.

BACKGROUND: Online clinical risk prediction tools built on data from multiple cohorts are increasingly being utilized for contemporary doctor-patient decision-making and validation. This report outlines a comprehensive data science strategy for building such tools with application to the Prostate Biopsy Collaborative Group prostate cancer risk prediction tool. METHODS: We created models for high-grade prostate cancer risk using six established risk factors. The data comprised 8492 prostate biopsies collected from ten institutions, 2 in Europe and 8 across North America. We calculated area under the receiver operating characteristic curve (AUC) for discrimination, the Hosmer-Lemeshow test statistic (HLS) for calibration and the clinical net benefit at risk threshold 15%. We implemented several internal cross-validation schemes to assess the influence of modeling method and individual cohort on validation performance. RESULTS: High-grade disease prevalence ranged from 18% in Zurich (1863 biopsies) to 39% in UT Health San Antonio (899 biopsies). Visualization revealed outliers in terms of risk factors, including San Juan VA (51% abnormal digital rectal exam), Durham VA (63% African American), and Zurich (2.8% family history). Exclusion of any cohort did not significantly affect the AUC or HLS, nor did the choice of prediction model (pooled, random-effects, meta-analysis). Excluding the lowest-prevalence Zurich cohort from training sets did not statistically significantly change the validation metrics for any of the individual cohorts, except for Sunnybrook, where the effect on the AUC was minimal. Therefore the final multivariable logistic model was built by pooling the data from all cohorts using logistic regression. Higher prostate-specific antigen and age, abnormal digital rectal exam, African ancestry and a family history of prostate cancer increased risk of high-grade prostate cancer, while a history of a prior negative prostate biopsy decreased risk (all p-values < 0.004). CONCLUSIONS: We have outlined a multi-cohort model-building internal validation strategy for developing globally accessible and scalable risk prediction tools.

Association between variants in genes involved in the immune response and prostate cancer risk in men randomized to the finasteride arm in the Prostate Cancer Prevention Trial.

BACKGROUND: We reported that some, but not all single nucleotide polymorphisms (SNPs) in select immune response genes are associated with prostate cancer, but not individually with the prevalence of intraprostatic inflammation in the Prostate Cancer Prevention Trial (PCPT) placebo arm. Here, we investigated whether these same SNPs are associated with risk of lower- and higher-grade prostate cancer in men randomized to finasteride, and with prevalence of intraprostatic inflammation among controls. Methods A total of 16 candidate SNPs in IL1ß, IL2, IL4, IL6, IL8, IL10, IL12(p40), IFNG, MSR1, RNASEL, TLR4, and TNFA and 7 tagSNPs in IL10 were genotyped in 625 white prostate cancer cases, and 532 white controls negative for cancer on an end-of-study biopsy nested in the PCPT finasteride arm. We used logistic regression to estimate log-additive odds ratios (OR) and 95% confidence intervals (CI) adjusting for age and family history. RESULTS: Minor alleles of rs2243250 (T) in IL4 (OR = 1.46, 95% CI 1.03-2.08, P-trend = 0.03), rs1800896 (G) in IL10 (OR = 0.77, 95% CI 0.61-0.96, P-trend = 0.02), rs2430561 (A) in IFNG (OR = 1.33, 95% CI 1.02-1.74; P-trend = 0.04), rs3747531 (C) in MSR1 (OR = 0.55, 95% CI 0.32-0.95; P-trend = 0.03), and possibly rs4073 (A) in IL8 (OR = 0.81, 95% CI 0.64-1.01, P-trend = 0.06) were associated with higher- (Gleason 7-10; N = 222), but not lower- (Gleason 2-6; N = 380) grade prostate cancer. In men with low PSA (<2 ng/mL), these higher-grade disease associations were attenuated and/or no longer significant, whereas associations with higher-grade disease were apparent for minor alleles of rs1800795 (C: OR = 0.70, 95% CI 0.51-0.94, P-trend = 0.02) and rs1800797 (A: OR = 0.72, 95% CI 0.53-0.98, P-trend = 0.04) in IL6. While some IL10 tagSNPs were associated with lower- and higher-grade prostate cancer, distributions of IL10 haplotypes did not differ, except possibly between higher-grade cases and controls among those with low PSA (P = 0.07). We did not observe an association between the studied SNPs and intraprostatic inflammation in the controls. CONCLUSION: In the PCPT finasteride arm, variation in genes involved in the immune response, including possibly IL8 and IL10 as in the placebo arm, may be associated with prostate cancer, especially higher-grade disease, but not with intraprostatic inflammation. We cannot rule out PSA-associated detection bias or chance due to multiple testing.

MBDDiff: an R package designed specifically for processing MBDcap-seq datasets.

BACKGROUND: Since its initial discovery in 1975, DNA methylation has been intensively studied and shown to be involved in various biological processes, such as development, aging and tumor progression. Many experimental techniques have been developed to measure the level of DNA methylation. Methyl-CpG binding domain-based capture followed by high-throughput sequencing (MBDCap-seq) is a widely used method for characterizing DNA methylation patterns in a genome-wide manner. However, current methods for processing MBDCap-seq datasets does not take into account of the region-specific genomic characteristics that might have an impact on the measurements of the amount of methylated DNA (signal) and background fluctuation (noise). Thus, specific software needs to be developed for MBDCap-seq experiments. RESULTS: A new differential methylation quantification algorithm for MBDCap-seq, MBDDiff, was implemented. To evaluate the performance of the MBDDiff algorithm, a set of simulated signal based on negative binomial and Poisson distribution with parameters estimated from real MBDCap-seq datasets accompanied with different background noises were generated, and then performed against a set of commonly used algorithms for MBDCap-seq data analysis in terms of area under the ROC curve (AUC), number of false discoveries and statistical power. In addition, we also demonstrated the effective of MBDDiff algorithm to a set of in-house prostate cancer samples, endometrial cancer samples published earlier, and a set of public-domain triple negative breast cancer samples to identify potential factors that contribute to cancer development and recurrence. CONCLUSIONS: In this paper we developed an algorithm, MBDDiff, designed specifically for datasets derived from MBDCap-seq. MBDDiff contains three modules: quality assessment of datasets and quantification of DNA methylation; determination of differential methylation of promoter regions; and visualization functionalities. Simulation results suggest that MBDDiff performs better compared to MEDIPS and DESeq in terms of AUC and the number of false discoveries at different levels of background noise. MBDDiff outperforms MEDIPS with increased backgrounds noise, but comparable performance when noise level is lower. By applying MBDDiff to several MBDCap-seq datasets, we were able to identify potential targets that contribute to the corresponding biological processes. Taken together, MBDDiff provides user an accurate differential methylation analysis for data generated by MBDCap-seq, especially under noisy conditions.