Detection and Grading of Prostate Cancer in Temporal Enhanced Ultrasound using Convolutional Neural Networks

Prostate cancer is the most common non-dermatological cancer amongst men in the developed world. The current definitive diagnosis is core needle biopsy guided by transrectal ultrasound. However, this method suffers from low sensitivity and specificity in detecting cancer. Recently, a new ultrasound based tissue typing approach has been proposed, known as temporal enhanced ultrasound (TeUS). In this approach, a set of temporal ultrasound frames is collected from a stationary tissue location without any intentional mechanical excitation. The main aim of this thesis is to implement a deep learning-based solution for prostate cancer detection and grading using TeUS data. In the proposed solution, convolutional neural networks are trained to extract high-level features from time domain TeUS data in temporally and spatially adjacent frames in nine in vivo prostatectomy cases. This approach avoids information loss due to feature extraction and also improves cancer detection rate. The output likelihoods of two TeUS arrangements are then combined to form our novel decision support system. This deep learning-based approach results in the area under the receiver operating characteristic curve (AUC) of 0.80 and 0.73 for prostate cancer detection and grading, respectively, in leave-one-patient-out cross-validation. Recently, multi-parametric magnetic resonance imaging (mp-MRI) has been utilized to improve detection rate of aggressive prostate cancer. In this thesis, for the first time, we present the fusion of mp-MRI and TeUS for characterization of prostate cancer to compensates the deficiencies of each image modalities and improve cancer detection rate. The results obtained using TeUS are fused with those attained using consolidated mp-MRI maps from multiple MR modalities and cancer delineations on those by multiple clinicians. The proposed fusion approach yields the AUC of 0.86 in prostate cancer detection. The outcomes of this thesis emphasize the viable potential of TeUS as a tissue typing method. Employing this ultrasound-based intervention, which is non-invasive and inexpensive, can be a valuable and practical addition to enhance the current prostate cancer detection.

What Was I Thinking? Eye-Tracking Experiments Underscore the Bias that Architecture Exerts on Nuclear Grading in Prostate Cancer.

We previously reported that nuclear grade assignment of prostate carcinomas is subject to a cognitive bias induced by the tumor architecture. Here, we asked whether this bias is mediated by the non-conscious selection of nuclei that "match the expectation" induced by the inadvertent glance at the tumor architecture. 20 pathologists were asked to grade nuclei in high power fields of 20 prostate carcinomas displayed on a computer screen. Unknown to the pathologists, each carcinoma was shown twice, once before a background of a low grade, tubule-rich carcinoma and once before the background of a high grade, solid carcinoma. Eye tracking allowed to identify which nuclei the pathologists fixated during the 8 second projection period. For all 20 pathologists, nuclear grade assignment was significantly biased by tumor architecture. Pathologists tended to fixate on bigger, darker, and more irregular nuclei when those were projected before kigh grade, solid carcinomas than before low grade, tubule-rich carcinomas (and vice versa). However, the morphometric differences of the selected nuclei accounted for only 11% of the architecture-induced bias, suggesting that it can only to a small part be explained by the unconscious fixation on nuclei that "match the expectation". In conclusion, selection of « matching nuclei » represents an unconscious effort to vindicate the gravitation of nuclear grades towards the tumor architecture.

What was I thinking? Eye-tracking experiments underscore the bias that architecture exerts on nuclear grading in prostate cancer

We previously reported that nuclear grade assignment of prostate carcinomas is subject to a cognitive bias induced by the tumor architecture. Here, we asked whether this bias is mediated by the non-conscious selection of nuclei that "match the expectation" induced by the inadvertent glance at the tumor architecture. 20 pathologists were asked to grade nuclei in high power fields of 20 prostate carcinomas displayed on a computer screen. Unknown to the pathologists, each carcinoma was shown twice, once before a background of a low grade, tubule-rich carcinoma and once before the background of a high grade, solid carcinoma. Eye tracking allowed to identify which nuclei the pathologists fixated during the 8 second projection period. For all 20 pathologists, nuclear grade assignment was significantly biased by tumor architecture. Pathologists tended to fixate on bigger, darker, and more irregular nuclei when those were projected before kigh grade, solid carcinomas than before low grade, tubule-rich carcinomas (and vice versa). However, the morphometric differences of the selected nuclei accounted for only 11% of the architecture-induced bias, suggesting that it can only to a small part be explained by the unconscious fixation on nuclei that "match the expectation". In conclusion, selection of « matching nuclei » represents an unconscious effort to vindicate the gravitation of nuclear grades towards the tumor architecture.

Biopsy diagnosis of intraductal carcinoma is prognostic in intermediate and high risk prostate cancer patients treated by radiotherapy.

AIM: We investigated the prognostic significance of intraductal carcinoma of the prostate (IDC-P) in biopsies and transurethral resections prior to external beam radiotherapy with or without androgen deprivation. METHODS: Cohort 1 consisted of 118 intermediate risk prostate cancer patients treated by radiotherapy, with biochemical relapse as primary end-point (median follow-up 6.5 years). Cohort 2 consisted of 132 high risk patients, enrolled in a phase III randomised trial (EORTC 22863) comparing radiotherapy alone to radiotherapy with long-term androgen deprivation (LTAD) with clinical progression free survival as primary end-point (median follow-up 9.1 years). Presence of IDC-P was identified after central review. Multivariable regression modelling and Kaplan-Meier analysis were performed with IDC-P as dichotomous variable. RESULTS: IDC-P was a strong prognosticator for early (<36 months) biochemical relapse (HR 7.3; p = 0.007) in cohort 1 and for clinical disease-free survival in both arms of cohort 2 (radiotherapy arm: HR 3.5; p < 0.0001; radiotherapy plus LTAD arm: HR 2.8, p = 0.018). IDC-P retained significance after stratification for reviewed Gleason score in the radiotherapy arm (HR 2.3; p = 0.03). IDC-P was a strong prognosticator for metastatic failure rate (radiotherapy arm: HR 5.3; p < 0.0001; radiotherapy plus LTAD arm: HR 3.6; p = 0.05). CONCLUSIONS: IDC-P in diagnostic samples of patients with intermediate or high risk prostate cancer is an independent prognosticator of early biochemical relapse and metastatic failure rate after radiotherapy. We suggest that the presence of IDC-P in prostate biopsies should routinely be reported.

Clinical intervals and diagnostic characteristics in a cohort of prostate cancer patients in Spain: a multicentre observational study.

BACKGROUND Little is known about the healthcare process for patients with prostate cancer, mainly because hospital-based data are not routinely published. The main objective of this study was to determine the clinical characteristics of prostate cancer patients, the, diagnostic process and the factors that might influence intervals from consultation to diagnosis and from diagnosis to treatment. METHODS We conducted a multicentre, cohort study in seven hospitals in Spain. Patients' characteristics and diagnostic and therapeutic variables were obtained from hospital records and patients' structured interviews from October 2010 to September 2011. We used a multilevel logistic regression model to examine the association between patient care intervals and various variables influencing these intervals (age, BMI, educational level, ECOG, first specialist consultation, tumour stage, PSA, Gleason score, and presence of symptoms) and calculated the odds ratio (OR) and the interquartile range (IQR). To estimate the random inter-hospital variability, we used the median odds ratio (MOR). RESULTS 470 patients with prostate cancer were included. Mean age was 67.8 (SD: 7.6) years and 75.4 % were physically active. Tumour size was classified as T1 in 41.0 % and as T2 in 40 % of patients, their median Gleason score was 6.0 (IQR:1.0), and 36.1 % had low risk cancer according to the D'Amico classification. The median interval between first consultation and diagnosis was 89 days (IQR:123.5) with no statistically significant variability between centres. Presence of symptoms was associated with a significantly longer interval between first consultation and diagnosis than no symptoms (OR:1.93, 95%CI 1.29-2.89). The median time between diagnosis and first treatment (therapeutic interval) was 75.0 days (IQR:78.0) and significant variability between centres was found (MOR:2.16, 95%CI 1.45-4.87). This interval was shorter in patients with a high PSA value (p = 0.012) and a high Gleason score (p = 0.026). CONCLUSIONS Most incident prostate cancer patients in Spain are diagnosed at an early stage of an adenocarcinoma. The period to complete the diagnostic process is approximately three months whereas the therapeutic intervals vary among centres and are shorter for patients with a worse prognosis. The presence of prostatic symptoms, PSA level, and Gleason score influence all the clinical intervals differently.

Prognostic role of genetic biomarkers in clinical progression of prostate cancer.

The aim of this study was to analyze the use of 12 single-nucleotide polymorphisms in genes ELAC2, RNASEL and MSR1 as biomarkers for prostate cancer (PCa) detection and progression, as well as perform a genetic classification of high-risk patients. A cohort of 451 men (235 patients and 216 controls) was studied. We calculated means of regression analysis using clinical values (stage, prostate-specific antigen, Gleason score and progression) in patients and controls at the basal stage and after a follow-up of 72 months. Significantly different allele frequencies between patients and controls were observed for rs1904577 and rs918 (MSR1 gene) and for rs17552022 and rs5030739 (ELAC2). We found evidence of increased risk for PCa in rs486907 and rs2127565 in variants AA and CC, respectively. In addition, rs627928 (TT-GT), rs486907 (AG) and rs3747531 (CG-CC) were associated with low tumor aggressiveness. Some had a weak linkage, such as rs1904577 and rs2127565, rs4792311 and rs17552022, and rs1904577 and rs918. Our study provides the proof-of-principle that some of the genetic variants (such as rs486907, rs627928 and rs2127565) in genes RNASEL, MSR1 and ELAC2 can be used as predictors of aggressiveness and progression of PCa. In the future, clinical use of these biomarkers, in combination with current ones, could potentially reduce the rate of unnecessary biopsies and specific treatments.

p53 and cyclooxygenase-2 expression are directly associated with cyclin D1 expression in radical prostatectomy specimens of patients with hormone-naïve prostate cancer.

Prostate cancer (PCa) is a potentially curable disease when diagnosed in early stages and subsequently treated with radical prostatectomy (RP). However, a significant proportion of patients tend to relapse early, with the emergence of biochemical failure (BF) as an established precursor of progression to metastatic disease. Several candidate molecular markers have been studied in an effort to enhance the accuracy of existing predictive tools regarding the risk of BF after RP. We studied the immunohistochemical expression of p53, cyclooxygenase-2 (COX-2) and cyclin D1 in a cohort of 70 patients that underwent RP for early stage, hormone naïve PCa, with the aim of prospectively identifying any possible interrelations as well as correlations with known prognostic parameters such as Gleason score, pathological stage and time to prostate-specific antigen (PSA) relapse. We observed a significant (p = 0.003) prognostic role of p53, with high protein expression correlating with shorter time to BF (TTBF) in univariate analysis. Both p53 and COX-2 expression were directly associated with cyclin D1 expression (p = 0.055 and p = 0.050 respectively). High p53 expression was also found to be an independent prognostic factor (p = 0.023). Based on previous data and results provided by this study, p53 expression exerts an independent negative prognostic role in localized prostate cancer and could therefore be evaluated as a useful new molecular marker to be added in the set of known prognostic indicators of the disease. With respect to COX-2 and cyclin D1, further studies are required to elucidate their role in early prediction of PCa relapse after RP.

Histopathological characteristics of a novel knock-in mouse prostate cancer model

Prostate cancer is relatively unique to man. There is no naturally occurring prostate cancer in the mouse. Pre-clinical studies involve the establishment of a genetically engineered mouse prostate cancer model with features close to those of the human situation. A new knock-in mouse adenocarcinoma prostate (KIMAP) model was established, which showed close-to-human kinetics of tumor development. In order to determine if the similar kinetics is associated with heterogeneous tumor architecture similar to the human situation, we utilized a new mouse histological grading system (Gleason analogous grading system) similar to the Gleason human grading system and flow cytometry DNA analysis to measure and compare the adenocarcinoma of the KIMAP model with human prostate cancer. Sixty KIMAP prostate cancer samples from 60 mice were measured and compared with human prostate cancer. Flow cytometry DNA analysis was performed on malignant prostate tissues obtained from KIMAP models. Mice with prostate cancer from KIMAP models showed a 53.3% compound histological score rate, which was close to the human clinical average (50%) and showed a significant correlation with age (P = 0.001). Flow cytometry analyses demonstrated that most KIMAP tumor tissues were diploid, analogous to the human situation. The similarities of the KIMAP mouse model with tumors of the human prostate suggest the use of this experimental model to complement studies of human prostate cancer.

Correlation Between Testosterone And Psa Kinetics In Metastatic Prostate Cancer Patients Treated With Diverse Chemical Castrations.

To assess total testosterone and prostatic-specific antigen (PSA) kinetics among diverse chemical castrations, advanced-stage prostate cancer patients were randomized into three groups of 20: Group 1, Leuprolide 3.75 mg; Group 2, Leuprolide 7.5 mg; and Group 3, Goserelin 3.6 mg. All groups were treated with monthly application of the respective drugs. The patients' levels of serum total testosterone and PSA were evaluated at two time periods: before the treatment and 3 months after the treatment. Spearman's rank correlation coefficient was utilized to verify the hypothesis of linear correlation between total testosterone and PSA levels. At the beginning the patients' age, stage, grade, PSA, and total testosterone were similar within the three groups, with median age 72, 70, and 70 years in Groups 1, 2, and 3, respectively. Three months after the treatment, patients who received Leuprolide 7.5 mg presented significantly lower median total testosterone levels compared with Goserelin 3.6 mg and Leuprolide 3.75 mg (9.5 ng/dL vs. 20.0 ng/dL vs. 30.0 ng/dL, respectively; p = .0072), while those who received Goserelin 3.6 mg presented significantly lower PSA levels compared with Leuprolide 7.5 mg and Leuprolide 3.75 mg (0.67 vs. 1.86 vs. 2.57, respectively; p = .0067). There was no linear correlation between total testosterone and PSA levels. Overall, regarding castration levels of total testosterone, 28.77% of patients did not obtain levels ≤50 ng/dL and 47.80% did not obtain levels ≤20 ng/dL. There was no correlation between total testosterone and PSA kinetics and no equivalence among different pharmacological castrations.

Burden Of Focal Cryoablation Versus Brachytherapy Versus Active Surveillance In The Treatment Of Very Low-risk Prostate Cancer: A Preliminary Head-to-head Comprehensive Assessment.

Focal cryoablation (FC), brachytherapy (B) and active surveillance (AS) were offered to patients diagnosed with very low-risk prostate cancer (VLRPC) in an equal access protocol. Comprehensive validated self-report questionnaires accessed patients' erectile (IIEF-5) and voiding (IPSS) functions, Beck scales measured anxiety (BAI), hopelessness (BHS) and depression (BDI), SF-36 reflected patients' quality of life added to the emotional thermometers including five visual analogue scales (distress, anxiety, depression, anger and need for help). Kruskal-Wallis or ANOVA tests and Spearman's correlations were obtained among groups and studied variables. Thirty patients were included, median follow-up 18 months (15-21). Those on AS (n = 11) were older, presented higher hopelessness (BHS) and lower general health perceptions (SF-36) scores than patients opting for FC (n = 10) and B (n = 9), P = 0.0014, P = 0.0268 and P = 0.0168 respectively. Patients on B had higher IPSS scores compared to those under FC and AC, P = 0.0223. For all 30 included patients, Spearman's correlation (rs ) was very strong between BHS and general health perceptions (rs  = -0.800, P < 0.0001), and weak/moderate between age and BHS (rs  = 0.405, P = 0.026) and age and general health perceptions (rs  = -0.564, P = 0.001). The sample power was >60%. To be considered in patients' counselling and care, current study supports the hypothesis that even VLRPC when untreated undermines psychosocial domains.

Milk intake and the risk of type 2 diabetes mellitus, hypertension and prostate cancer

Milk intake is widely recommended for a healthy diet. Recent evidences suggest that milk/dairy products are associated with a lower risk of type 2 diabetes and hypertension. On the other hand, high calcium intake has been associated with a higher risk of prostate cancer. The calcium and vitamin D content in dairy foods could have beneficial effects on glucose metabolism and renin/angiotensin system as well regulates body weight. The association between high dairy/calcium consumption and prostate cancer risk are related to the presence of estrogens and insulin like growth factor (IGF-I) in milk. Based on the current evidence, it is possible that milk/dairy products, when consumed in adequate amounts and mainly with reduced fat content, has a beneficial effect on the prevention of hypertension and diabetes. Its potential role in the pathogenesis of prostate cancer is not well supported and requires additional study.

Lineage relationship of prostate cancer cell types based on gene expression

Background: Prostate tumor heterogeneity is a major factor in disease management. Heterogeneity could be due to multiple cancer cell types with distinct gene expression. Of clinical importance is the so-called cancer stem cell type. Cell type-specific transcriptomes are used to examine lineage relationship among cancer cell types and their expression similarity to normal cell types including stem/progenitor cells. Methods: Transcriptomes were determined by Affymetrix DNA array analysis for the following cell types. Putative prostate progenitor cell populations were characterized and isolated by expression of the membrane transporter ABCG2. Stem cells were represented by embryonic stem and embryonal carcinoma cells. The cancer cell types were Gleason pattern 3 (glandular histomorphology) and pattern 4 (aglandular) sorted from primary tumors, cultured prostate cancer cell lines originally established from metastatic lesions, xenografts LuCaP 35 (adenocarcinoma phenotype) and LuCaP 49 (neuroendocrine/small cell carcinoma) grown in mice. No detectable gene expression differences were detected among serial passages of the LuCaP xenografts. Results: Based on transcriptomes, the different cancer cell types could be clustered into a luminal-like grouping and a non-luminal-like (also not basal-like) grouping. The non-luminal-like types showed expression more similar to that of stem/progenitor cells than the luminal-like types. However, none showed expression of stem cell genes known to maintain stemness. Conclusions: Non-luminal-like types are all representatives of aggressive disease, and this could be attributed to the similarity in overall gene expression to stem and progenitor cell types.

Gene expression relationship between prostate cancer cells of Gleason 3, 4 and normal epithelial cells as revealed by cell type-specific transcriptomes

Background: Prostate cancer cells in primary tumors have been typed CD10(-)/CD13(-)/CD24(hi)/CD26(+)/CD38(lo)/CD44(-)/CD104(-). This CD phenotype suggests a lineage relationship between cancer cells and luminal cells. The Gleason grade of tumors is a descriptive of tumor glandular differentiation. Higher Gleason scores are associated with treatment failure. Methods: CD26(+) cancer cells were isolated from Gleason 3+3 (G3) and Gleason 4+4 (G4) tumors by cell sorting, and their gene expression or transcriptome was determined by Affymetrix DNA array analysis. Dataset analysis was used to determine gene expression similarities and differences between G3 and G4 as well as to prostate cancer cell lines and histologically normal prostate luminal cells. Results: The G3 and G4 transcriptomes were compared to those of prostatic cell types of non-cancer, which included luminal, basal, stromal fibromuscular, and endothelial. A principal components analysis of the various transcriptome datasets indicated a closer relationship between luminal and G3 than luminal and G4. Dataset comparison also showed that the cancer transcriptomes differed substantially from those of prostate cancer cell lines. Conclusions: Genes differentially expressed in cancer are potential biomarkers for cancer detection, and those differentially expressed between G3 and G4 are potential biomarkers for disease stratification given that G4 cancer is associated with poor outcomes. Differentially expressed genes likely contribute to the prostate cancer phenotype and constitute the signatures of these particular cancer cell types.

A Randomized, Double-Blind, Placebo-Controlled Phase II Study of Vandetanib Plus Docetaxel/Prednisolone in Patients with Hormone-Refractory Prostate Cancer

Vandetanib (ZACTIMA(TM)) is a once-daily oral anticancer drug that selectively inhibits vascular endothelial growth factor receptor, epidermal growth factor receptor, and rearranged during transfection signaling. This randomized (1: 1), double-blind study evaluated vandetanib (100mg/day) or placebo in combination with docetaxel (D; 75mg/m(2) every 3 weeks) and prednisolone (P; 2 x 5 mg/day) in 86 patients with metastatic hormone-refractory prostate cancer (mHRPC). The primary assessment was prostate-specific antigen (PSA) response (confirmed reduction of >= 50% from baseline) and a greater number of patients showed a PSA response with placebo + DP (67%) versus vandetanib + DP (40%); hazard ratio = 2.23 (one-sided 80% confidence limit = 2.90; one-sided p = 0.99). More patients experienced progression events (disease progression or death from any cause) with vandetanib + DP (65%) versus placebo + DP (60%); hazard ratio = 1.13 (one-sided 80% confidence limit = 1.44; one-sided p = 0.67). The overall incidence of adverse events was similar in both groups, although more patients experienced adverse events, leading to permanent discontinuation with vandetanib + DP (28%) versus placebo + DP (12%). However, the safety and tolerability profile for vandetanib was similar to that previously reported; adverse events that occurred more frequently in the vandetanib + DP arm were hypertension (14% vs. 2%), erythematous rash (14% vs. 2%), and exfoliative rash (12% vs. 2%). In this study of patients with mHRPC, vandetanib + DP did not demonstrate any efficacy benefit, compared with placebo + DP.