Dataset Information

Deep Learning with Quantitative Features of Magnetic Resonance Images to Predict Biochemical Recurrence of Radical Prostatectomy: A Multi-Center Study.

ABSTRACT: Biochemical recurrence (BCR) occurs in up to 27% of patients after radical prostatectomy (RP) and often compromises oncologic survival. To determine whether imaging signatures on clinical prostate magnetic resonance imaging (MRI) could noninvasively characterize biochemical recurrence and optimize treatment. We retrospectively enrolled 485 patients underwent RP from 2010 to 2017 in three institutions. Quantitative and interpretable features were extracted from T2 delineated tumors. Deep learning-based survival analysis was then applied to develop the deep-radiomic signature (DRS-BCR). The model's performance was further evaluated, in comparison with conventional clinical models. The model achieved C-index of 0.802 in both primary and validating cohorts, outweighed the CAPRA-S score (0.677), NCCN model (0.586) and Gleason grade group systems (0.583). With application analysis, DRS-BCR model can significantly reduce false-positive predictions, so that nearly one-third of patients could benefit from the model by avoiding overtreatments. The deep learning-based survival analysis assisted quantitative image features from MRI performed well in prediction for BCR and has significant potential in optimizing systemic neoadjuvant or adjuvant therapies for prostate cancer patients.

SUBMITTER: Yan Y

PROVIDER: S-EPMC8234539 | biostudies-literature | 2021 Jun

REPOSITORIES: biostudies-literature

ACCESS DATA

Publications

Deep Learning with Quantitative Features of Magnetic Resonance Images to Predict Biochemical Recurrence of Radical Prostatectomy: A Multi-Center Study.

Yan Ye Y Shao Lizhi L Liu Zhenyu Z He Wei W Yang Guanyu G Liu Jiangang J Xia Haizhui H Zhang Yuting Y Chen Huiying H Liu Cheng C Lu Min M Ma Lulin L Sun Kai K Zhou Xuezhi X Ye Xiongjun X Wang Lei L Tian Jie J Lu Jian J

Cancers 20210621 12

Biochemical recurrence (BCR) occurs in up to 27% of patients after radical prostatectomy (RP) and often compromises oncologic survival. To determine whether imaging signatures on clinical prostate magnetic resonance imaging (MRI) could noninvasively characterize biochemical recurrence and optimize treatment. We retrospectively enrolled 485 patients underwent RP from 2010 to 2017 in three institutions. Quantitative and interpretable features were extracted from T2 delineated tumors. Deep learning ...[more]

PMID: 34205786

Similar Datasets

Project description:The objective of this study was to utilize a microarray platform to determine the microRNA (miRNA) expression profile in prostate cancer tissue from patients who underwent radical prostatectomy and were followed for an average of 47 months. The GeneChip® miRNA Array (Affymetrix) contains more than 46,000 probes, and more than 1,100 mature human miRNAs and transcripts. An expression level higher or lower than 1.1 with a p<0.05 were considered to be differentially expressed. When tumors with a GS<7 were compared with those with a GS≥7, five miRNAs were down-regulated and nine were up-regulated. In comparing tumors from patients with PSA <10 ng/mL vs. those with PSA ≥10 ng/mL, 21 miRNAs were down-regulated and 29 were overexpressed. There was decreased expression of 21 miRNAs and increased expression of 22 miRNAs when pT2 and pT3 tumors were compared. The comparison of the expression profiles of men with vs. without recurrent disease (PSA>0.2 ng/mL) revealed that 14 miRNAs were down-regulated and 17 were up-regulated. Considering the three prognostic factors and biochemical recurrence, miR-335 was overexpressed in pT3 tumors in patients with PSA>10 ng/mL, miR-1202 was overexpressed in recurrent tumors in which PSA >10 ng/mL, and miR-885 and miR-1226 were overexpressed in recurrent pT3 tumors. Six miRNAs (miR- 361, 378, 548, 574, 652 and 892) had decreased expression in pT3 tumors in patients with PSA >10 ng/mL. miR-16 was down-regulated in recurrent pT3 tumors. miR-92 and miR-450 exhibited decreased expression in recurrent tumors in patients with PSA > 10 ng/mL. miR-28 was down-regulated in pT3 adenocarcinomas with a GS ≥ 7. The differentially expressed miRNAs in recurrent tumors were validated using qRT-PCR in 74 additional specimens. miR-21 was overexpressed in recurrent tumors, and in the Cox model, expression levels higher than 0.64 were indicative of biochemical recurrence with a risk of 2.4. These results, which were based on clinical specimens from patients treated in a standardized way and followed for an average of 47 months, provide an initial understanding of the role of miRNAs in prostatecancer progression; the foremost representative is miR-21, which has been implicated in the progression of various human cancers.

Project description:The objective of this study was to utilize a microarray platform to determine the microRNA (miRNA) expression profile in prostate cancer tissue from patients who underwent radical prostatectomy and were followed for an average of 47 months. The GeneChipM-BM-. miRNA Array (Affymetrix) contains more than 46,000 probes, and more than 1,100 mature human miRNAs and transcripts. An expression level higher or lower than 1.1 with a p<0.05 were considered to be differentially expressed. When tumors with a GS<7 were compared with those with a GSM-bM-^IM-%7, five miRNAs were down-regulated and nine were up-regulated. In comparing tumors from patients with PSA <10 ng/mL vs. those with PSA M-bM-^IM-%10 ng/mL, 21 miRNAs were down-regulated and 29 were overexpressed. There was decreased expression of 21 miRNAs and increased expression of 22 miRNAs when pT2 and pT3 tumors were compared. The comparison of the expression profiles of men with vs. without recurrent disease (PSA>0.2 ng/mL) revealed that 14 miRNAs were down-regulated and 17 were up-regulated. Considering the three prognostic factors and biochemical recurrence, miR-335 was overexpressed in pT3 tumors in patients with PSA>10 ng/mL, miR-1202 was overexpressed in recurrent tumors in which PSA >10 ng/mL, and miR-885 and miR-1226 were overexpressed in recurrent pT3 tumors. Six miRNAs (miR- 361, 378, 548, 574, 652 and 892) had decreased expression in pT3 tumors in patients with PSA >10 ng/mL. miR-16 was down-regulated in recurrent pT3 tumors. miR-92 and miR-450 exhibited decreased expression in recurrent tumors in patients with PSA > 10 ng/mL. miR-28 was down-regulated in pT3 adenocarcinomas with a GS M-bM-^IM-% 7. The differentially expressed miRNAs in recurrent tumors were validated using qRT-PCR in 74 additional specimens. miR-21 was overexpressed in recurrent tumors, and in the Cox model, expression levels higher than 0.64 were indicative of biochemical recurrence with a risk of 2.4. These results, which were based on clinical specimens from patients treated in a standardized way and followed for an average of 47 months, provide an initial understanding of the role of miRNAs in prostatecancer progression; the foremost representative is miR-21, which has been implicated in the progression of various human cancers. Fifty-three patients underwent a radical prostatectomy performed by the same surgeon (MS) to treat localized PC between August 2000 and June 2002. The mean age was 65 years (SD: 7.5, range: 49 - 77). The mean Gleason score was 7 (SD: 1.1, range: 5 - 10); 15 tumors had a score of 5 or 6, and 38 had a score between 7 and 10. Twenty-four (45.3%) of the patients had pT2 tumors, and 29 (54.7%) had pT3 tumors. The control group consisted of three patients with lower urinary tract symptoms, a mean age of 55.3 years and a diagnosis of benign prostatic hyperplasia (BPH) who consented to open surgery.

Project description:Accurate estimation of recurrence risk is needed for optimal treatment of clinically localized prostate cancer (PCa) patients. We combined classical clinicopathological predictors of biochemical recurrence (BCR) and molecular markers into a new risk score for predicting BCR after radical prostatectomy (RP). A retrospective study which included 122 PCa patients who attended our department between 2000 and 2007 was conducted. Total RNA was isolated from formalin-fixed paraffin embedded PCa tissue. Gene expression patterns were analyzed in 21 selected samples from 7 localized, 6 locally advanced, and 8 metastatic PCa patients using Illumina microarrays. Expression levels of 41 genes were validated by quantitative PCR in an independent retrospective series of 101 patients with ≥ 10 years follow-up who underwent RP for clinically localized PCa. Univariate and multivariate logistic regression analysis were used to identify individual predictors of BCR. A risk score (RS) for predicting BCR including clinicopathological and gene expression data was developed and Kaplan-Meier curves were generated. Interaction networks between the genes of the model were built by GeneMANIA Cytoscape plugin. In a median follow-up of 120 months (range 3-190), 37 patients developed BCR (36.6%). Expression levels of 7,930 transcripts differed between clinically localized and locally advanced-metastatic PCa groups (FDR<0.1). A set of 41 genes were validated by quantitative PCR. Regression analysis showed that expression of ASF1B and MCL1 as well as Gleason score, extracapsular extension, seminal vesicle invasion and positive margins were independent prognostic factors of BCR. A RS generated using these gene expression and clinicopathological variables was able to discriminate between two groups of patients with a significantly different probability of BCR (HR 6.24; CI 3.23-12.4, p<0.01), improving the individual discriminative performance of each of these variables on their own. Direct interactions between the two genes of the model were not found. In conclusion, identification of new gene expression patterns associated with a high probability of BCR in clinically localized PCa patients treated with RP, and their combination with clinicopathological variables in a robust, easy-to-use and reliable classifier may contribute to improve PCa risk stratification and, consequently, to tailor treatment and surveillance strategies in these patients.

Dataset Information

Deep Learning with Quantitative Features of Magnetic Resonance Images to Predict Biochemical Recurrence of Radical Prostatectomy: A Multi-Center Study.

Publications

Deep Learning with Quantitative Features of Magnetic Resonance Images to Predict Biochemical Recurrence of Radical Prostatectomy: A Multi-Center Study.

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure

Tweets