PURPOSE Prostate-specific antigen testing has led to overtreatment of prostate cancer (PCa). Only a small subset of PCa patients will have an aggressive disease that requires intensive therapy, and there is currently no biomarker to predict disease aggressiveness at the time of surgery. MicroRNAs (miRNAs) are reported to be involved in PCa pathogenesis. METHODS This study involved 105 participants. For the discovery phase, prostatectomy samples were dichotomized to high-risk (n = 27, biochemical failure <36 months after prostatectomy) and low-risk groups (n = 14, ≥36 months without biochemical failure). Expression of 754 mature miRNAs was compared between the 2 groups. Linear regression models were built to accurately predict biochemical failure risk. miRNA mimics were transfected into PCa model cell lines to test effects on proliferation and to deduce responding signaling pathways. RESULTS We identified 25 differentially expressed miRNAs between the biochemical failure risk groups. Based on the expression of 2–3 miRNAs, 3 logistic regression models were developed, each with a high positive predictive value. Candidate miRNAs and the best-performing model were also verified on an independent PCa set. miRNA-152, featured in the models, was further investigated by using cell line models and was shown to affect cell proliferation. Predicted interaction between miR-152 and (mRNA)ERBB3 (erythroblastic leukemia viral oncogene homolog 3) was experimentally validated in vitro. CONCLUSIONS miRNAs can help to predict biochemical failure risk at the time of prostatectomy.

Purpose: Forkhead box Q1 (FOXQ1) has been shown to contribute to the development and progression of cancers, including ovarian and breast cancer (BC). However, research exploring FOXQ1 expression, copy number variation (CNV), and prognostic value across different BC subtypes is limited. Our purpose was to evaluate FOXQ1 mRNA expression, CNV, and prognostic value across BC subtypes. Materials and Methods: We determined FOXQ1 expression and CNV in BC patient tumors using RT-qPCR and qPCR, respectively. We also analyzed FOXQ1 expression and CNV in BC cell lines in the CCLE database using K-means clustering. The prognostic value of FOXQ1 expression in the TCGA-BRCA database was assessed using univariate and multivariate Cox's regression analysis as well as using the online tools OncoLnc, GEPIA, and UALCAN. Results: Our analyses reveal that FOXQ1 mRNA is differentially expressed between different subtypes of BC and is significantly decreased in luminal BC and HER2 patients when compared to normal breast tissue samples. Furthermore, analysis of BC cell lines showed that FOXQ1 mRNA expression was independent of CNV. Moreover, patients with low FOXQ1 mRNA expression had significantly poorer overall survival compared to those with high FOXQ1 mRNA expression. Finally, low FOXQ1 expression had a critical impact on the prognostic values of BC patients and was an independent predictor of overall survival when it was adjusted for BC subtypes and to two other FOX genes, FOXF2 and FOXM1. Conclusion: Our study reveals for the first time that FOXQ1 is differentially expressed across BC subtypes and that low expression of FOXQ1 is indicative of poor prognosis in patients with BC.

As a member of the forkhead box (FOX) superfamily of transcription factors, FOXQ1 plays a critical role in a wide range of biological processes, including angiogenesis, epithelial differentiation and smooth muscle differentiation. Emerging evidence also show FOXQ1 to play important roles in the development and progression of various cancers such as ovarian, pancreatic, and colorectal cancer. Particularly, FOXQ1 has been linked to facilitating tumor invasion and metastasis in breast cancer and has also been specifically associated with Triple Negative Basal‐like Breast Cancer (TN/BL BC). Furthermore, experimental studies have demonstrated FOXQ1 overexpression to promote/mediate epithelial to mesenchymal transition, invasion, stemness traits, and chemoresistance in breast cancer. These processes contribute substantially to poor prognosis in breast cancer patients, thus making FOXQ1 a potential target for the diagnosis and treatment of breast cancer tumors. In this study, we investigated the mechanisms leading to increased expression levels of FOXQ1 in BC through analysis of FOXQ1 copy number variation (CNV) and mRNA levels across BC patient subtypes and cell lines. Additionally, we assessed the prognostic significance of FOXQ1 in BC patients. Finally, K‐means clustering was conducted by using Python coding to identify unique clusters for FOXQ1 mRNA levels and CNV in BC cell lines. We report for the first time, that FOXQ1 mRNA is differentially expressed across BC patients. FOXQ1 mRNA is significantly down regulated in Luminal (ER+) BC patients (n=6) when compared to control samples (n=6). FOXQ1mRNA expression is significantly up regulated in TN/BL BC patients (n=6) compared to Her2 (n=6) and ER+ BC (n=6) respectively. We also found FOXQ1 significantly has more copies in TN/BL BC compared to control samples. K‐means clustering analysis was conducted on BC cell lines, with the purpose of identifying distinct subpopulations based on FOXQ1 mRNA expression and CNV. Our supervised and unsupervised clustering analyses identified 3 and 4 clusters, respectively, among BC cell lines for FOXQ1 mRNA compared to their CNV. However, we found FOXQ1 mRNA expression to be independent of its CNV. Moreover, FOXQ1 mRNA was found to be highly expressed in numerous TN/BL cell lines. Lastly and most importantly, by applying the bioinformatic online tool GEPIA, Kaplan‐Meier survival curve analysis identified two risk groups with high (n=524) and low (n=531) FOXQ1 mRNA expression levels. Our Kaplan‐Meier survival analysis also showed patients with low FOXQ1 mRNA expression to have shorter overall survival time than those with high FOXQ1 mRNA expression (HR=0.71, P=0.042). In accordance to these results, we propose that FOXQ1 can serve as an emerging new prognostic biomarker for BC. Understanding the mechanism(s) underlying FOXQ1’s activation in breast cancer could facilitate the development of improved therapies for BC patients. Support or Funding Information Women and Children Health Research Institute Faculty of Medicine and Dentistry, U...

194 Background: With the introduction of PSA testing, the problem of over-treatment emerged in prostate cancer. Only a small subset of prostate cancer patients will require more intensive adjuvant therapy. There is currently no biomarker that can predict disease aggressiveness at the time of surgery. Methods: We analyzed miRNA expression in 41 patients (the discovery set) which were dichotomized into; 'high risk'- experienced biochemical failure within 24 months after radical prostatectomy (n=26) and 'low risk' who did not have biochemical failure for at least 35 months (n=15). The validation set consisted of 72 cases. Total RNA was isolated from FFPE cores. cDNA was prepared for each patients and expression miRNA expression was screened by qRT-PCR –based panel. miRNAs were ranked by non-parametric tests. Linear regression models were built to predict biochemical failure. We used TargetScan for miRNA target prediction. Targets were validated by transient transfection of synthetic miRNA precursors followed by qRT-PCR quantification of the targets. Proliferation was assessed by measuring cell viability. Results: We compared the expression of 754 mature human miRNAs in patients with ‘high’ or ‘low’ risk for biochemical failure. We identified 24 miRNAs that were differentially expressed between the risk groups. We developed three logistic regression models, based on the expression of 2-3 miRNAs (PPV=100% and NPV ranges 86.4-100%). We confirmed the differential expression on the study set and on a larger, independent set of PCa pateints. We also validated one model on an independent set of patients. Further, we show that transfection of miR-152 and miR-331-3p, featured in the logistic regression models, altered proliferation of PCa3 and DU145 cells. Target prediction indicated Erbb3 and Erbb2 as potential direct targets and their mRNA expression significantly reduced when miR-152 and miR-331-3p were overexpressed. Conclusions: Altered miR-331-3p and miR-152 expression represent a potential tool for assessing the risk of early biochemical failure. These miRNAs may act through the Erbb family to induce an alternative way of AR activation.

10622 Background: In rodents, 10-20% of the genome has a 24-hour (h) rhythm in RNA expression. A molecular clock consisting of transcription / translation feedback loops of clock-genes controls this rhythmicity. In a microarray study (Affymetrix HG_U133_Plus2 chip) on RNA extracted from CLL cells sampled every 4 hours over 24 hours (6 samples) from 3 male (M) and 6 female (F) patients (pts) with chronic lymphocytic leukemia (CLL) we found 15,094 and 13,415 rhythmic transcripts (Cosinor analysis) in M and F respectively, only 6,629 of which were common to both M and F. We hypothesized that these gender differences in rhythmic RNA expression might have clinical implication for biomarker discovery with some important biomarkers only found at a certain times of day and with gender differences. Methods: Sampling was performed again at 6 time points in M and F CLL patients with indolent CLL (10pts, 5 M, 5F) and aggressive CLL (10 pts, 5M, 5F). An 8-plex iTRAQ kit was employed for mass spectrometry (MS) based relative quantification. Each iTRAQ block comprised all time-points for a single patient including a universal control. Results: At least 300 proteins were identified at a 95% confidence level in each iTRAQ experiment. Of these, 74 proteins displayed significant change between aggressive and indolent pts in 24-hour average or single time point normalized expression based on a non-parametric U-test (p<0.05). The JTK_CYCLE algorithm was used to detect 24-hour rhythmic patterns in the proteomic datasets. Significant rhythmic expression was found for 44 proteins (p<0.10) identified by iTRAQ. Data for each time point were independently analyzed using an in-house Butterfly clustering algorithm based on discrete dynamical systems. The 4 PM time point was found to be optimal for stratifying aggressive and indolent disease. Cellular pathways with significant association to differentially expressed proteins included PPAR signaling, fatty acid metabolism, and granzyme-A signaling. Confirmation by selected reaction monitoring (SRM) MS is ongoing. Conclusions: Rhythmic protein expression may have clinical implications for biomarker discovery.