Transcriptome-wide prediction of prostate cancer gene expression from histopathology images using co-expression-based convolutional neural networks.

MOTIVATION: Molecular phenotyping by gene expression profiling is central in contemporary cancer research and in molecular diagnostics but remains resource intense to implement. Changes in gene expression occurring in tumours cause morphological changes in tissue, which can be observed on the microscopic level. The relationship between morphological patterns and some of the molecular phenotypes can be exploited to predict molecular phenotypes from routine haematoxylin and eosin-stained whole slide images (WSIs) using convolutional neural networks (CNNs). In this study, we propose a new, computationally efficient approach to model relationships between morphology and gene expression. RESULTS: We conducted the first transcriptome-wide analysis in prostate cancer, using CNNs to predict bulk RNA-sequencing estimates from WSIs for 370 patients from the TCGA PRAD study. Out of 15 586 protein coding transcripts, 6618 had predicted expression significantly associated with RNA-seq estimates (FDR-adjusted P-value <1×10-4) in a cross-validation and 5419 (81.9%) of these associations were subsequently validated in a held-out test set. We furthermore predicted the prognostic cell-cycle progression score directly from WSIs. These findings suggest that contemporary computer vision models offer an inexpensive and scalable solution for prediction of gene expression phenotypes directly from WSIs, providing opportunity for cost-effective large-scale research studies and molecular diagnostics. AVAILABILITY AND IMPLEMENTATION: A self-contained example is available from http://github.com/phiwei/prostate_coexpression. Model predictions and metrics are available from doi.org/10.5281/zenodo.4739097. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

Incorporating Magnetic Resonance Imaging and Biomarkers in Active Surveillance Protocols - Results From the Prospective Stockholm3 Active Surveillance Trial (STHLM3AS).

BACKGROUND: Active surveillance (AS) for men with low-risk prostate cancer (PC) can lead to patient morbidity and healthcare overutilization. The aim of this study was to evaluate an AS protocol using the Stockholm3 test and magnetic resonance imaging (MRI) to reduce biopsy intensity. METHODS: We conducted a prospective multicenter study of 280 invited men from a contemporary screening study (STHLM3), with Gleason Score (GS) 3 + 3 PC on a current AS protocol. Patients underwent prostate-MRI and blood sampling for analysis of the Stockholm3 test including protein biomarkers, genetic variants, and clinical variables to predict risk of GS ≥3 + 4 PC followed by systematic biopsies and targeted biopsies (for Prostate Imaging Reporting and Data System version 2 ≥3 lesions) in all men. Primary outcomes were reclassification to GS ≥3 + 4 PC and clinically significant PC (csPCa), including unfavorable intermediate risk PC or higher based on National Comprehensive Cancer Network guidelines. RESULTS: Adding MRI-targeted biopsies to systematic biopsies increased sensitivity of GS ≥3 + 4 PC compared with systematic biopsies alone (relative sensitivity [RS] = 1.52, 95% confidence interval [CI] = 1.28 to 1.85). Performing biopsies in only MRI positive increased sensitivity of GS ≥3 + 4 PC (RS = 1.30, 95% CI = 1.04 to 1.67) and reduced number of biopsy procedures by 49.3% while missing 7.2% GS ≥3 + 4 PC and 1.4% csPCa. Excluding men with negative Stockholm3 test reduced the number of MRI investigations at follow-up by 22.5% and biopsies by 56.8% while missing 6.9% GS ≥3 + 4 PC and 1.3% csPCa. CONCLUSION: Including MRI and targeted/systematic biopsies in the follow-up for men on AS increased sensitivity of PC reclassification. Incorporation of risk prediction models including biomarkers may reduce the need for MRI use in men with low-risk PC.

Polygenic hazard score to guide screening for aggressive prostate cancer: development and validation in large scale cohorts.

OBJECTIVES: To develop and validate a genetic tool to predict age of onset of aggressive prostate cancer (PCa) and to guide decisions of who to screen and at what age. DESIGN: Analysis of genotype, PCa status, and age to select single nucleotide polymorphisms (SNPs) associated with diagnosis. These polymorphisms were incorporated into a survival analysis to estimate their effects on age at diagnosis of aggressive PCa (that is, not eligible for surveillance according to National Comprehensive Cancer Network guidelines; any of Gleason score ≥7, stage T3-T4, PSA (prostate specific antigen) concentration ≥10 ng/L, nodal metastasis, distant metastasis). The resulting polygenic hazard score is an assessment of individual genetic risk. The final model was applied to an independent dataset containing genotype and PSA screening data. The hazard score was calculated for these men to test prediction of survival free from PCa. SETTING: Multiple institutions that were members of international PRACTICAL consortium. PARTICIPANTS: All consortium participants of European ancestry with known age, PCa status, and quality assured custom (iCOGS) array genotype data. The development dataset comprised 31 747 men; the validation dataset comprised 6411 men. MAIN OUTCOME MEASURES: Prediction with hazard score of age of onset of aggressive cancer in validation set. RESULTS: In the independent validation set, the hazard score calculated from 54 single nucleotide polymorphisms was a highly significant predictor of age at diagnosis of aggressive cancer (z=11.2, P<10-16). When men in the validation set with high scores (>98th centile) were compared with those with average scores (30th-70th centile), the hazard ratio for aggressive cancer was 2.9 (95% confidence interval 2.4 to 3.4). Inclusion of family history in a combined model did not improve prediction of onset of aggressive PCa (P=0.59), and polygenic hazard score performance remained high when family history was accounted for. Additionally, the positive predictive value of PSA screening for aggressive PCa was increased with increasing polygenic hazard score. CONCLUSIONS: Polygenic hazard scores can be used for personalised genetic risk estimates that can predict for age at onset of aggressive PCa.

Investigating the possible causal role of coffee consumption with prostate cancer risk and progression using Mendelian randomization analysis.

Coffee consumption has been shown in some studies to be associated with lower risk of prostate cancer. However, it is unclear if this association is causal or due to confounding or reverse causality. We conducted a Mendelian randomisation analysis to investigate the causal effects of coffee consumption on prostate cancer risk and progression. We used two genetic variants robustly associated with caffeine intake (rs4410790 and rs2472297) as proxies for coffee consumption in a sample of 46,687 men of European ancestry from 25 studies in the PRACTICAL consortium. Associations between genetic variants and prostate cancer case status, stage and grade were assessed by logistic regression and with all-cause and prostate cancer-specific mortality using Cox proportional hazards regression. There was no clear evidence that a genetic risk score combining rs4410790 and rs2472297 was associated with prostate cancer risk (OR per additional coffee increasing allele: 1.01, 95% CI: 0.98,1.03) or having high-grade compared to low-grade disease (OR: 1.01, 95% CI: 0.97,1.04). There was some evidence that the genetic risk score was associated with higher odds of having nonlocalised compared to localised stage disease (OR: 1.03, 95% CI: 1.01, 1.06). Amongst men with prostate cancer, there was no clear association between the genetic risk score and all-cause mortality (HR: 1.00, 95% CI: 0.97,1.04) or prostate cancer-specific mortality (HR: 1.03, 95% CI: 0.98,1.08). These results, which should have less bias from confounding than observational estimates, are not consistent with a substantial effect of coffee consumption on reducing prostate cancer incidence or progression.

Total antioxidant intake and prostate cancer in the Cancer of the Prostate in Sweden (CAPS) study. A case control study.

BACKGROUND: The total intake of dietary antioxidants may reduce prostate cancer risk but available data are sparse and the possible role of supplements unclear. We investigated the potential association between total and dietary antioxidant intake and prostate cancer in a Swedish population. METHODS: We used FFQ data from 1499 cases and 1112 controls in the population based case-control study Cancer of the Prostate in Sweden (CAPS). The ferric reducing antioxidant potential (FRAP) assay was used to assess the total antioxidant capacity (TAC) of diet and supplements. We calculated odds ratios (ORs) for the risk of prostate cancer across quintiles of antioxidant intake from all foods, from fruit and vegetables only, and from dietary supplements using unconditional logistic regression. RESULTS: Coffee comprised 62 % of the dietary antioxidant intake, tea 4 %, berries 4 %, chocolate 2 %, and boiled potatoes 2 %. In total 19 % and 13 % of the population took multivitamins and supplemental Vitamin C respectively, on a regular basis. Antioxidant intake from all foods and from fruits and vegetables separately measured by the FRAP assay was not associated with prostate cancer risk. For antioxidant intake from supplements we found a positive association with total, advanced, localized, high grade and low grade prostate cancer in those above median supplemental TAC intake of users compared to non-users (Adjusted ORs for total prostate cancer: 1.37, 95 % CI 1.08-1.73, advanced: 1.51, 95 % CI 1.11-2.06, localized: 1.36. 95 % CI 1.06-1.76, high grade 1.60, 95 % CI 1.06-2.40, low grade 1.36, 95 % CI 1.03-1.81). A high intake of coffee (≥6 cups/day) was associated with a possible risk reduction of fatal and significantly with reduced risk for high grade prostate cancer, adjusted OR: 0.45 (95 % CI: 0.22-0.90), whereas a high intake of chocolate was positively associated with risk of total, advanced, localized and low grade disease (adjusted OR for total: 1.43, 95 % CI 1.12-1.82, advanced: 1.40, 95 % CI 1.01-1.96, localized: 1.43, 95 % CI 1.08-1.88, low-grade: 1.41, 95 % CI 1.03-1.93). CONCLUSIONS: Total antioxidant intake from diet was not associated with prostate cancer risk. Supplement use may be associated with greater risk of disease.

Coffee and risk of prostate cancer incidence and mortality in the Cancer of the Prostate in Sweden Study.

PURPOSE: Coffee intake has recently been associated with significantly lower risk of lethal and advanced prostate cancer in a US population. METHODS: We studied the association between coffee and prostate cancer risk in the population-based case-control study Cancer of the Prostate in Sweden. Dietary data were available for 1,499 cases and 1,112 controls. We calculated odds ratios (ORs) for the risk of prostate cancer in high versus low categories of coffee intake using logistic regression. We studied overall prostate cancer risk as well as risk of fatal, advanced, localized, high-grade, grade 7, and low-grade disease. RESULTS: Mean coffee intake was 3.1 cups per day among both cases and controls. Coffee intake was not associated with overall prostate cancer risk. Risk of fatal prostate cancer was inversely, but not statistically significantly, associated with coffee intake, with an odds ratio of 0.64 [95 % confidence interval (CI) 0.34-1.19, p value for linear trend = 0.81] for men consuming greater than 5 cups per day compared to men drinking less than 1 cup per day. The highest intake of coffee was associated non-significantly with lower risk of advanced disease (OR = 0.73, 95 % CI 0.41-1.30, p trend = 0.98) and associated significantly with lower risk of high-grade cancer (Gleason 8-10; OR = 0.50, 95 % CI 0.26-0.98, p trend = 0.13). Risk of localized, grade 7, and low-grade cancers was not associated with coffee intake. CONCLUSIONS: This study provides some support of an inverse association between coffee and lethal and high-grade prostate cancer.

Interaction between single nucleotide polymorphisms in selenoprotein P and mitochondrial superoxide dismutase determines prostate cancer risk.

Selenium may affect prostate cancer risk via its plasma carrier selenoprotein P which shows dramatically reduced expression in prostate cancer tumors and cell lines. The selenoprotein P (SEPP1) Ala234 single nucleotide polymorphism (SNP) allele is associated with lower plasma selenoprotein P in men, reducing the concentration/activity of other antioxidant selenoproteins. Selenium status also modifies the effect of the mitochondrial superoxide dismutase (SOD2) SNP Ala16Val on prostate cancer risk. We investigated the relationship of these SNPs with prostate cancer risk. DNA from 2,975 cases and 1,896 age-matched controls from the population-based Prostate Cancer in Sweden study were genotyped using TaqMan assays. Cases were designated aggressive or nonaggressive prostate cancers at diagnosis by clinical criteria. Association with prostate cancer was investigated by logistic regression; gene-gene interaction using a general linear model. The mean plasma selenium concentration measured in 169 controls was relatively low (76.0 +/- 17.2 microg/L). SNP genotype distributions were in Hardy-Weinberg equilibrium. SOD2-Ala16+ men were at a greater risk of prostate cancer [odds ratios (OR), 1.19; 95% confidence intervals (CI), 1.03-1.37] compared with SOD2-Val16 homozygotes. Men homozygous for SEPP1-Ala234 who were also SOD2-Ala16+ had a higher risk of prostate cancer (OR, 1.43; 95% CI, 1.17-1.76) and aggressive prostate cancer (OR, 1.60; 95% CI, 1.22-2.09) than those who were SOD2-Val16 homozygotes (interaction, prostate cancer P = 0.05; aggressive prostate cancer P = 0.01). This interaction was stronger in ever-smokers: SOD2-Ala16+ men homozygous for SEPP1-Ala234 had an almost doubled risk of prostate cancer (OR, 1.97; 95% CI, 1.33-2.91; interaction P = 0.001). In a low-selenium population, SOD2-Ala16+ men homozygous for SEPP1-Ala234 are at an increased risk of prostate cancer/aggressive prostate cancer especially if ever-smokers, because they are likely to produce more mitochondrial H(2)O(2) that they cannot remove, thereby promoting prostate tumor cell proliferation and migration.

Association of frequent consumption of fatty fish with prostate cancer risk is modified by COX-2 polymorphism.

Dietary intake of marine fatty acids from fish may protect against prostate cancer development. We studied this association and whether it is modified by genetic variation in cyclooxygenase (COX)-2, a key enzyme in fatty acid metabolism and inflammation. We assessed dietary intake of fish among 1,499 incident prostate cancer cases and 1,130 population controls in Sweden. Five single nucleotide polymorphisms (SNPs) were identified and genotyped in available blood samples for 1,378 cases and 782 controls. Odds ratios (OR) and 95% confidence intervals (CI) were estimated by multivariate logistic regression. Multiplicative and additive interactions between fish intake and COX-2 SNPs on prostate cancer risk were evaluated. Eating fatty fish (e.g., salmon-type fish) once or more per week, compared to never, was associated with reduced risk of prostate cancer (OR: 0.57, 95% CI: 0.43-0.76). The OR comparing the highest to the lowest quartile of marine fatty acids intake was 0.70 (95% CI: 0.51-0.97). We found a significant interaction (p < 0.001) between salmon-type fish intake and a SNP in the COX-2 gene (rs5275: +6365 T/C), but not with the 4 other SNPs examined. We found strong inverse associations with increasing intake of salmon-type fish among carriers of the variant allele (OR for once per week or more vs. never = 0.28, 95% CI: 0.18-0.45; p(trend) < 0.01), but no association among carriers of the more common allele. Frequent consumption of fatty fish and marine fatty acids appears to reduce the risk of prostate cancer, and this association is modified by genetic variation in the COX-2 gene.

Dietary intake of phytoestrogens, estrogen receptor-beta polymorphisms and the risk of prostate cancer.

BACKGROUND: The causes of prostate cancer are poorly understood, but genetic factors may be more important than for many other malignancies, and dietary phytoestrogens may be protective. Because phytoestrogens bind tightly to the estrogen receptor-beta, we conducted an epidemiologic investigation of synergistic effects between phytoestrogen intake and estrogen receptor-beta gene polymorphisms. METHODS: We performed a population-based case-control study in Sweden. All participants reported their phytoestrogen intake and donated a blood sample. We identified four haplotype-tagging single nucleotide polymorphisms (htSNPs) and genotyped these htSNPs in 1314 prostate cancer patients and 782 controls. Odds ratios were estimated by multivariate logistic regression. Interactions between phytoestrogen intake and estrogen receptor-beta SNPs on prostate cancer risk were evaluated considering both multiplicative and additive effect scales. RESULTS: We found a significant multiplicative interaction (P = 0.04) between dietary intake of phytoestrogens and a promoter SNP in the estrogen receptor-beta gene (rs 2987983-13950), but not with any of the three other htSNPs (P = 0.11, 0.69, 0.85). Among carriers of the variant promoter alleles, we found strong inverse associations with increasing intake of total phytoestrogens (odds ratio for highest vs. lowest quartile = 0.43; P for trend <0.001), isoflavonoids (odds ratio = 0.63; P for trend = 0.05), and coumestrol (odds ratio = 0.57; P for trend = 0.003). We found no association between phytoestrogens and prostate cancer among carriers homozygous for the wild-type allele (TT). CONCLUSIONS: Our study provides strong evidence that high intake of phytoestrogens substantially reduce prostate cancer risk among men with specific polymorphic variation in the promoter region of the estrogen receptor-beta gene.

Dietary phytoestrogen, serum enterolactone and risk of prostate cancer: the cancer prostate Sweden study (Sweden).

OBJECTIVE: Based on evidence that phytoestrogens may protect against prostate cancer, we evaluated the associations between serum enterolactone concentration or dietary phytoestrogen intake and risk of prostate cancer. METHODS: In our Swedish population-based case-control study, questionnaire-data were available for 1,499 prostate cancer cases and 1,130 controls, with serum enterolactone levels in a sub-group of 209 cases and 214 controls. Unconditional logistic regression was performed to estimate multivariate odds ratios (ORs) and 95% confidence intervals (CIs) for associations with risk of prostate cancer. RESULTS: High intake of food items rich in phytoestrogens was associated with a decreased risk of prostate cancer. The OR comparing the highest to the lowest quartile of intake was 0.74 (95% CI: 0.57-0.95; p-value for trend: 0.01). In contrast, we found no association between dietary intake of total or individual lignans or isoflavonoids and risk of prostate cancer. Intermediate serum levels of enterolactone were associated with a decreased risk of prostate cancer. The ORs comparing increasing quartiles of serum enterolactone concentration to the lowest quartile were, respectively, 0.28 (95% CI: 0.15-0.55), 0.63 (95% CI: 0.35-1.14) and 0.74 (95% CI: 0.41-1.32). CONCLUSIONS: Our results support the hypothesis that certain foods high in phytoestrogens are associated with a lower risk of prostate cancer.

Adjuvant chemotherapy in colorectal cancer: a joint analysis of randomised trials by the Nordic Gastrointestinal Tumour Adjuvant Therapy Group.

Due to uncertainties regarding clinically meaningful gains from adjuvant chemotherapy after colorectal cancer surgery, several Nordic Groups in the early 1990s initiated randomised trials to prove or reject such gains. This report gives the joint analyses after a minimum 5-year follow-up. Between October 1991 and December 1997, 2 224 patients under 76 years of age with colorectal cancer stages II and III were randomised to surgery alone (n = 1 121) or adjuvant chemotherapy (n = 1 103) which varied between trials (5FU/levamisole for 12 months, n = 444; 5FU/leucovorin for 4-5 months according to either a modified Mayo Clinic schedule (n = 262) or the Nordic schedule (n = 397). Some centres also randomised patients treated with 5FU/leucovorin to+/-levamisole). A total of 812 patients had colon cancer stage II, 708 colon cancer stage III, 323 rectal cancer stage II and 368 rectal cancer stage III. All analyses were according to intention-to-treat. No statistically significant difference in overall survival, stratified for country or region, could be found in any group of patients according to stage or site. In colon cancer stage III, an absolute difference of 7% (p = 0.15), favouring chemotherapy, was seen. The present analyses corroborate a small but clinically meaningful survival gain from adjuvant chemotherapy in colon cancer stage III, but not in the other presentations.

Prostate cancer epidemiology.

Because more and more men are being diagnosed with prostate cancer worldwide, knowledge about and prevention of this disease is important. Epidemiological studies have provided some insight about the cause of prostate cancer in terms of diet and genetic factors. However, compared with other common cancers such as breast and lung cancer, the causes remain poorly understood. Several important issues could help in our understanding of this disease-the variation in incidence of prostate cancer between ethnic populations and the factors leading to familial clustering of the diseases.