scholarly journals Predicting prostate cancer specific-mortality with artificial intelligence-based Gleason grading

2021 ◽  
Vol 1 (1) ◽  
Author(s):  
Ellery Wulczyn ◽  
Kunal Nagpal ◽  
Matthew Symonds ◽  
Melissa Moran ◽  
Markus Plass ◽  
...  
Keyword(s):  
Prostate Cancer ◽  
Artificial Intelligence ◽  
Risk Groups ◽  
Risk Scores ◽  
Pathology Report ◽  
Gleason Grading ◽  
Specific Mortality ◽  
Cancer Specific Mortality ◽  
Open Question

Abstract Background Gleason grading of prostate cancer is an important prognostic factor, but suffers from poor reproducibility, particularly among non-subspecialist pathologists. Although artificial intelligence (A.I.) tools have demonstrated Gleason grading on-par with expert pathologists, it remains an open question whether and to what extent A.I. grading translates to better prognostication. Methods In this study, we developed a system to predict prostate cancer-specific mortality via A.I.-based Gleason grading and subsequently evaluated its ability to risk-stratify patients on an independent retrospective cohort of 2807 prostatectomy cases from a single European center with 5–25 years of follow-up (median: 13, interquartile range 9–17). Results Here, we show that the A.I.’s risk scores produced a C-index of 0.84 (95% CI 0.80–0.87) for prostate cancer-specific mortality. Upon discretizing these risk scores into risk groups analogous to pathologist Grade Groups (GG), the A.I. has a C-index of 0.82 (95% CI 0.78–0.85). On the subset of cases with a GG provided in the original pathology report (n = 1517), the A.I.’s C-indices are 0.87 and 0.85 for continuous and discrete grading, respectively, compared to 0.79 (95% CI 0.71–0.86) for GG obtained from the reports. These represent improvements of 0.08 (95% CI 0.01–0.15) and 0.07 (95% CI 0.00–0.14), respectively. Conclusions Our results suggest that A.I.-based Gleason grading can lead to effective risk stratification, and warrants further evaluation for improving disease management.

The association of very low PSA with increased cancer-specific death in men with high-grade prostate cancer.

2015 ◽  
Vol 33 (7_suppl) ◽  
pp. 62-62
Author(s):  
Brandon Arvin Virgil Mahal ◽  
Ayal Aaron Aizer ◽  
Jason Alexander Efstathiou ◽  
Paul Linh Nguyen
Keyword(s):  
Prostate Cancer ◽  
Reference Group ◽  
High Grade ◽  
Specific Mortality ◽  
National Cohort ◽  
Poorly Differentiated ◽  
Cancer Specific Mortality ◽  
Prostate Cancers ◽  
Very High

62 Background: It has been hypothesized that very low PSAs in men with high-grade prostate cancer could reflect dedifferentiation and a poorer prognosis, but clinical evidence to support this is limited. We sought to determine whether a very low-presenting PSA was associated with greater prostate cancer-specific mortality (PCSM) among men with Gleason score (GS) 8-10 disease. Methods: The Surveillance, Epidemiology and End Results Program was used to identify a national cohort of 328,904 men diagnosed with cT1-4N0M0 prostate cancer between 2004 and 2010. Multivariable Fine-Gray competing-risks regression analysis was used to determine PCSM as a function of PSA level (<2.5 ng/mL, 2.6-4 ng/mL, 4.1-10 ng/mL, 10.1-20 ng/mL, 20.1-40 ng/mL, or >40ng/mL) and GS (8-10 vs. <=7). Results: Median follow-up was 38 months. Among men with GS 8-10 disease, using PSA 4.1-10 as the reference group, the Adjusted HR (AHR) for PCSM for men with PSA level <2.5 was 1.86 (95% CI 1.51-2.29; P<0.001), PSA 2.6-4 was1.44 (1.17-1.78; P<0.001), PSA 10.1-20 was 1.58 (1.39-1.78; P<0.001), PSA 20.1-40 was 2.04 (1.78-2.33; P<0.001), and PSA>40 was 3.19 (2.83-3.59; P<0.001), suggesting a U-shaped distribution. There was a significant interaction between PSA level and GS (Pinteraction<0.001) such that PSA <2.5 only significantly predicted for poorer PCSM among patients with high grade GS 8-10 disease. Conclusions: Among patients with high grade GS 8-10 disease, patients with PSA <2.5 and 2.6-4 appear to have a higher risk for cancer-specific death compared to patients with a 10.1-20 PSA level, supporting the notion that low PSA in GS 8-10 disease may be a sign of underlying aggressive and extremely poorly differentiated or anaplastic low PSA-producing tumors. Patients with low PSA GS 8-10 disease should be considered for clinical trials studying the use of chemotherapy and other novel agents in very-high risk prostate cancers.

Cancer-Specific Mortality After Surgery or Radiation for Patients With Clinically Localized Prostate Cancer Managed During the Prostate-Specific Antigen Era

2003 ◽  
Vol 21 (11) ◽  
pp. 2163-2172 ◽  
Author(s):  
Anthony V. D’Amico ◽  
Judd Moul ◽  
Peter R. Carroll ◽  
Leon Sun ◽  
Deborah Lubeck ◽  
...  
Keyword(s):  
Prostate Cancer ◽  
Prostate Specific Antigen ◽  
Specific Antigen ◽  
Risk Groups ◽  
The United States ◽  
Localized Prostate Cancer ◽  
Institutional Setting ◽  
Psa Failure ◽  
Specific Mortality ◽  
Cancer Specific Mortality

Purpose: To determine whether pretreatment risk groups shown to predict time to prostate cancer–specific mortality (PCSM) after treatment at a single institution retained that ability in a multi-institutional setting. Patients and Methods: From 1988 to 2002, 7,316 patients treated in the United States at 44 institutions with either surgery (n = 4,946) or radiation (n = 2,370) for clinical stage T1c-2, N0 or NX, M0 prostate cancer made up the study cohort. A Cox regression analysis was performed to determine the ability of pretreatment risk groups to predict time to PCSM after treatment. The relative risk (RR) of PCSM and 95% confidence intervals (CIs) were calculated for the intermediate- and high-risk groups relative to the low-risk group. Results: Estimates of non-PCSM 8 years after prostate-specific antigen (PSA) failure were 4% v 15% (surgery versus radiation; Plog rank = .002) compared with 13% v 18% (surgery versus radiation; Plog rank = .35) for patients whose age at the time of PSA failure was less than 70 as compared with ≥ 70 years, respectively. The RR of PCSM after treatment for surgery-managed patients with high- or intermediate-risk disease was 14.2 (95% CI, 5.0 to 23.4; PCox < .0001) and 4.9 (95% CI, 1.7 to 8.1; PCox = .0037), respectively. These values were 14.3 (95% CI, 5.2 to 24.0; PCox < .0001) and 5.6 (95% CI, 2.0 to 9.3; PCox = .0012) for radiation-managed patients. Conclusion: This study provided evidence to support the prediction of time to PCSM after surgery or radiation on the basis of pretreatment risk groups for patients with clinically localized prostate cancer managed during the PSA era.

Daily Aspirin Use and Prostate Cancer–Specific Mortality in a Large Cohort of Men with Nonmetastatic Prostate Cancer

2014 ◽  
Vol 32 (33) ◽  
pp. 3716-3722 ◽  
Author(s):  
Eric J. Jacobs ◽  
Christina C. Newton ◽  
Victoria L. Stevens ◽  
Peter T. Campbell ◽  
Stephen J. Freedland ◽  
...  
Keyword(s):  
Prostate Cancer ◽  
High Risk ◽  
Recent Analysis ◽  
Clinical Database ◽  
Specific Mortality ◽  
Cancer Specific Mortality ◽  
Daily Aspirin ◽  
High Doses ◽  
Large Prospective Cohort

Purpose In a recent analysis of a large clinical database, postdiagnosis aspirin use was associated with 57% lower prostate cancer–specific mortality (PCSM) among men diagnosed with nonmetastatic prostate cancer. However, information on this association remains limited. We assessed the association between daily aspirin use and PCSM in a large prospective cohort. Patients and Methods This analysis included men diagnosed with nonmetastatic prostate cancer between enrollment in the Cancer Prevention Study-II Nutrition Cohort in 1992 or 1993 and June 2009. Aspirin use was reported at enrollment, in 1997, and every 2 years thereafter. During follow-up through 2010, there were 441 prostate cancer deaths among 8,427 prostate cancer cases with information on prediagnosis aspirin use and 301 prostate cancer deaths among 7,118 prostate cancer cases with information on postdiagnosis aspirin use. Results Compared with no aspirin use, neither prediagnosis nor postdiagnosis daily aspirin use were statistically significantly associated with PCSM (prediagnosis use, multivariable-adjusted hazard ratio (HR) = 0.92, 95% CI 0.72 to 1.17, postdiagnosis use, HR = 0.98; 95% CI, 0.74 to 1.29). However, among men diagnosed with high-risk cancers (≥ T3 and/or Gleason score ≥ 8), postdiagnosis daily aspirin use was associated with lower PCSM (HR = 0.60; 95% CI, 0.37 to 0.97), with no clear difference by dose (low-dose, typically 81 mg per day, HR = 0.50; 95% CI, 0.27 to 0.92, higher dose, HR = 0.73; 95% CI, 0.40 to 1.34). Conclusion A randomized trial of aspirin among men diagnosed with nonmetastatic prostate cancer was recently funded. Our results suggest any additional randomized trials addressing this question should prioritize enrolling men with high-risk cancers and need not use high doses.

The Association between pre-operative PSA and prostate cancer-specific mortality in patients with long-term follow-up after radical prostatectomy

The Prostate ◽  
2011 ◽  
Vol 72 (1) ◽  
pp. 24-29 ◽  
Author(s):  
Dan Lewinshtein ◽  
Brandon Teng ◽  
Ashley Valencia ◽  
Robert Gibbons ◽  
Christopher R. Porter
Keyword(s):  
Prostate Cancer ◽  
Radical Prostatectomy ◽  
Specific Mortality ◽  
Long Term Follow Up ◽  
Cancer Specific Mortality

scholarly journals Prospective Study of Prostate Tumor Angiogenesis and Cancer-Specific Mortality in the Health Professionals Follow-Up Study

2009 ◽  
Vol 27 (33) ◽  
pp. 5627-5633 ◽  
Author(s):  
Lorelei A. Mucci ◽  
Anna Powolny ◽  
Edward Giovannucci ◽  
Zhiming Liao ◽  
Stacey A. Kenfield ◽  
...  
Keyword(s):  
Prostate Cancer ◽  
Microvessel Density ◽  
Health Professionals ◽  
Prostate Cancer Mortality ◽  
Follow Up Study ◽  
Vessel Lumen ◽  
Specific Mortality ◽  
Cancer Specific Mortality ◽  
And Function

Purpose Tumor growth requires the development of independent vascular networks that are often primitive in morphology and function. We examined whether microvessel morphology contributes to the considerable biologic heterogeneity of prostate cancer. Methods We evaluated microvessel morphology as a predictor of prostate cancer mortality among 572 men in the Health Professionals Follow-Up Study diagnosed with cancer during 1986 to 2000. We immunostained prostatectomy tumor block sections for endothelial marker CD34 and assessed microvessel density, vessel size (area and diameter), and irregularity of vessel lumen using image analysis. Proportional hazards models were used to assess microvessel density and morphology in relation to lethal prostate cancer. Results Poorly differentiated tumors exhibited greater microvessel density, greater irregularity of the vessel lumen, and smaller vessels. During 20 years of follow-up, 44 men developed bone metastases or died of cancer. Men with tumors exhibiting the smallest vessel diameter, based on quartiles, were 6.0 times more likely (95% CI, 1.8 to 20.0) to develop lethal prostate cancer. Men with the most irregularly shaped vessels were 17.1 times more likely (95% CI, 2.3 to 128) to develop lethal disease. Adjusting for Gleason grade and prostate-specific antigen levels did not qualitatively change the results. Microvessel density was not linked to cancer-specific mortality after adjusting for clinical factors. Conclusion Aggressive tumors form vessels that are primitive in morphology and function, with consequences for metastases. Vascular size and irregularity reflect the angiogenic potential of prostate cancer and may serve as biomarkers to predict prostate cancer mortality several years after diagnosis.

Association of expression of ERG oncoprotein with development of disease recurrence and prostate cancer-specific mortality.

2012 ◽  
Vol 30 (15_suppl) ◽  
pp. e15207-e15207
Author(s):  
Richard Johnston ◽  
Elysia Sophia Spencer ◽  
Xiaoyu Qu ◽  
Cigdem Himmetoglu Ussakli ◽  
Ryan Robert Gordon ◽  
...  
Keyword(s):  
Prostate Cancer ◽  
Recurrent Disease ◽  
Medical Center ◽  
Disease Recurrence ◽  
Patient Treatment ◽  
Genomic Alteration ◽  
Expression Intensity ◽  
Specific Mortality ◽  
Cancer Specific Mortality

e15207 Background: ETS-Related Gene (ERG) protein is present in approximately half of prostate cancer (PCa) specimens and correlates with TMPRSS2-ERG rearrangement. ERG oncogenic activation is believed to be a causal genomic alteration in a significant portion of PCa, its prognostic value in disease progression remains uncertain. This study evaluated ERG levels at radical prostatectomy (RP) to determine if it was predictive of earlier relapse or prostate cancer-specific mortality (PCSM). Methods: From 1954-1997, 1004 consecutive patients underwent RP at Virginia Mason Medical Center in Seattle. Recurrence was confirmed by tissue diagnosis or radiographic signs. Death certificates confirmed PCSM. Patients without tissue were excluded. 34 patients with metastases or PCSM were matched to patients without recurrence at a 1:2 ratio. Paraffin embedded tissue was stained with two anti-ERG monoclonal antibodies, clone 9FY (Furustao et al., PCPD, 2010) and clone EPR3864 (Epitomics, Burlingame, CA). A pathologist evaluated the nuclear expression intensity using a 4-point scale. Statistical analysis was performed on SAS. Results: Mean follow up was 10.26 yrs.Antibody 9FY detected ERG in 10/16 (62.5%) PCSM and 24/44 (54.6%) no recurrence patients. Cancer expression intensity was higher from patients that developed recurrent disease, 0.99 vs 1.59 (p = 0.0265) and PCSM 2.09 (p = 0.0002). Antibody EPR3864 detected ERG in 11/17 (64.7%) PCSM and 26/48 (52.1%) no recurrence patients. Cancer expression intensity was higher from patients that developed recurrent disease, 0.97 vs 1.64 (p = 0.011) and PCSM, 2.30 (p < 0.0001). The two antibodies were independent of PSA and grade and were highly correlated with each other (p < 0.0001). Conclusions: ERG expression at time of surgery was prognostic for the development of disease recurrence and PCSM. Previous studies have shown an increased lifetime risk of PCSM with ERG expression in a surveillance cohort. This is the first report linking PCSM to ERG expression in men treated with surgery. ERG scoring may be a useful metric to identify patients who would benefit from adjuvant treatment or closer follow-up, allowing more accurate individual patient treatment plans.

Prognostic utility of CCP score in men with prostate cancer after primary external beam radiation therapy.

2013 ◽  
Vol 31 (6_suppl) ◽  
pp. 47-47
Author(s):  
Stephen J. Freedland ◽  
Leah Gerber ◽  
Julia E. Reid ◽  
William Welbourn ◽  
Eliso Tikishvili ◽  
...  
Keyword(s):  
Prostate Cancer ◽  
Radiation Therapy ◽  
Localized Prostate Cancer ◽  
P Value ◽  
Clinical Parameters ◽  
Prognostic Information ◽  
Beam Radiation ◽  
Specific Mortality ◽  
Cancer Specific Mortality

47 Background: Accurate risk stratification improves decision making in localized prostate cancer. The CCP score, a prognostic RNA signature based on the average expression level of 31 cell cycle progression (CCP) genes, was developed to aid in this task. Previously, the CCP score was shown to be predictive of biochemical recurrence (BCR) after prostatectomy, and prostate cancer specific mortality in men undergoing observation. However, the value of CCP score in men undergoing primary electron beam radiation therapy (EBRT) was untested. Methods: The CCP score was derived retrospectively from the diagnostic biopsy of 141 patients treated with EBRT at the Durham VA medical Center. Inclusion criteria were disease diagnosis from 1991 to 2006 and available biopsy tissue. Approximately half of the cohort was African-American. Outcome was time from EBRT to BCR using Phoenix definition, and median follow-up for patients without BCR was 4.8 years. Association with outcome was evaluated by CoxPH survival analysis and likelihood ratio tests. Results: Patient data were censored at 5-years of follow-up and 19 patients (13%) had BCR. The median CCP score was 0.12(IQR -0.43 to 0.66). In univariable analysis, CCP score was a significant prognostic variable (p-value = 0.0017). The hazard ratio (HR) for BCR was 2.55 (95% CI (1.43, 4.55)) for a one-unit increase in CCP score (equivalent to a doubling of gene expression). In a predefined multivariable analysis that included Gleason score, PSA, and percent positive cores, the HR for CCP changed only marginally and remained significant (HR per CCP unit 2.09 (95% CI (1.05, 4.18), p-value = 0.035), indicating that CCP provides prognostic information that is not provided by standard clinical parameters. Further adjustment for hormonal therapy or radiation dose did not materially alter this association. The score was also associated with prostate cancer specific mortality. There was no evidence for interaction between CCP and any clinical variable, including ethnicity. Conclusions: In a cohort of men treated with EBRT, the CCP score was significantly associated with outcome and provided prognostic information beyond what was available from clinical parameters. If validated in a larger cohort, CCP score could be used to select high-risk men undergoing EBRT who may need combination therapy for their clinically localized prostate cancer.

Sign in / Sign up

Export Citation Format

Share Document