Automating Ground Truth Annotations for Gland Segmentation Through Immunohistochemistry.

Microscopic evaluation of glands in the colon is of utmost importance in the diagnosis of inflammatory bowel disease and cancer. When properly trained, deep learning pipelines can provide a systematic, reproducible, and quantitative assessment of disease-related changes in glandular tissue architecture. The training and testing of deep learning models require large amounts of manual annotations, which are difficult, time-consuming, and expensive to obtain. Here, we propose a method for automated generation of ground truth in digital hematoxylin and eosin (H&E)-stained slides using immunohistochemistry (IHC) labels. The image processing pipeline generates annotations of glands in H&E histopathology images from colon biopsy specimens by transfer of gland masks from KRT8/18, CDX2, or EPCAM IHC. The IHC gland outlines are transferred to coregistered H&E images for training of deep learning models. We compared the performance of the deep learning models to that of manual annotations using an internal held-out set of biopsy specimens as well as 2 public data sets. Our results show that EPCAM IHC provides gland outlines that closely match manual gland annotations (Dice = 0.89) and are resilient to damage by inflammation. In addition, we propose a simple data sampling technique that allows models trained on data from several sources to be adapted to a new data source using just a few newly annotated samples. The best performing models achieved average Dice scores of 0.902 and 0.89 on Gland Segmentation and Colorectal Adenocarcinoma Gland colon cancer public data sets, respectively, when trained with only 10% of annotated cases from either public cohort. Altogether, the performances of our models indicate that automated annotations using cell type-specific IHC markers can safely replace manual annotations. Automated IHC labels from single-institution cohorts can be combined with small numbers of hand-annotated cases from multi-institutional cohorts to train models that generalize well to diverse data sources.

The Clinical Cell-Cycle Risk (CCR) Score Is Associated With Metastasis After Radiation Therapy and Provides Guidance on When to Forgo Combined Androgen Deprivation Therapy With Dose-Escalated Radiation.

PURPOSE: The clinical cell-cycle risk (CCR) score, which combines the University of California, San Francisco's Cancer of the Prostate Risk Assessment (CAPRA) and the cell cycle progression (CCP) molecular score, has been validated to be prognostic of disease progression for men with prostate cancer. This study evaluated the ability of the CCR score to prognosticate the risk of metastasis in men receiving dose-escalated radiation therapy (RT) with or without androgen deprivation therapy (ADT). METHODS AND MATERIALS: This retrospective, multi-institutional cohort study included men with localized National Comprehensive Cancer Network (NCCN) intermediate-, high-, and very high-risk prostate cancer (N = 741). Patients were treated with dose-escalated RT with or without ADT. The primary outcome was time to metastasis. RESULTS: The CCR score prognosticated metastasis with a hazard ratio (HR) per unit score of 2.22 (95% confidence interval [CI], 1.71-2.89; P < .001). The CCR score better prognosticated metastasis than NCCN risk group (CCR, P < .001; NCCN, P = .46), CAPRA score (CCR, P = .002; CAPRA, P = .59), or CCP score (CCR, P < .001; CCP, P = .59) alone. In bivariable analyses, CCR score remained highly prognostic when accounting for ADT versus no ADT (HR, 2.18; 95% CI, 1.61-2.96; P < .001), ADT duration as a continuous variable (HR, 2.11; 95% CI, 1.59-2.79; P < .001), or ADT given at or below the recommended duration for each NCCN risk group (HR, 2.19; 95% CI, 1.69-2.86; P < .001). Men with CCR scores below or above the multimodality threshold (CCR score, 2.112) had a 10-year risk of metastasis of 3.7% and 21.24%, respectively. Men with below-threshold scores receiving RT alone had a 10-year risk of metastasis of 3.7%, and for men receiving RT plus ADT, the 10-year risk of metastasis was also 3.7%. CONCLUSIONS: The CCR score accurately and precisely prognosticates metastasis and adds clinically actionable information relative to guideline-recommended therapies based on NCCN risk in men undergoing dose-escalated RT with or without ADT. For men with scores below the multimodality threshold, adding ADT may not significantly reduce their 10-year risk of metastasis.

Validation of the cell cycle progression score to differentiate indolent from aggressive prostate cancer in men diagnosed through transurethral resection of the prostate biopsy.

BACKGROUND: Validation of biomarker-based prognostic models to improve risk stratification in men with localized prostate cancer (PrCa) remains a clinical need. It has previously been shown that the cell cycle progression (CCP) test provides significant, independent prognostic information for men who were incidentally diagnosed with PrCa after transurethral resection of the prostate (TURP) and were conservatively managed. AIM: The results have been extended in a newly analyzed retrospective cohort of UK men diagnosed through TURP biopsy (TURP1B; N = 305). METHODS AND RESULTS: The CCP score was derived from TURP biopsy tissue and combined with a modified UCSF Cancer of the Prostate Risk Assessment score (CAPRA) to generate the clinical cell-cycle risk score (CCR). The primary endpoint was PrCa-specific mortality (PSM). Hazard ratios (HR) were calculated for a one-unit change in score. Median follow-up was 9.6 (IQR: 5.4, 14.1) years, and 67 (22%) men died from PrCa within 10 years of diagnosis. The median CCP score was 1.1 (IQR: 0.6, 1.7). In univariate analyses, CCR proved a significant prognosticator of PSM (HR per unit score change = 2.28; 95% CI: 1.89, 2.74; P = 1.0 × 10-19 ). In multivariate analyses, CCR remained a significant prognosticator of PSM after adjusting for CAPRA (HR per unit score change = 4.36; 95% CI: 2.65, 7.16; P = 1.3 × 10-8 ), indicating that its molecular component, CCP, provides significant, independent prognostic information. CONCLUSION: These findings validate a combined clinicopathologic and molecular prognostic model for conservatively managed men who are diagnosed through TURP, supporting the use of CCR to inform clinical management.

Analysis of the prognostic utility of the cell cycle progression (CCP) score generated from needle biopsy in men treated with definitive therapy.

BACKGROUND: Accurate risk stratification can help guide appropriate treatment decisions in men with localized prostate cancer. Here, we evaluated the independent ability of the molecular cell cycle progression (CCP) score and the combined cell-cycle clinical risk (CCR) score to predict 10-year risk of progression to metastatic disease in a large, pooled analysis of men with definitively treated prostate cancer. METHODS: The pooled analysis included 1,062 patients from four institutions (Martini Clinic, Durham VA Medical Center, Intermountain Healthcare, Ochsner Clinic) treated definitively for localized prostate cancer by either radical prostatectomy or radiotherapy (brachytherapy or external beam radiotherapy ± hormone therapy). The CCP score was determined using the RNA expression of 46 genes from archival formalin-fixed paraffin-embedded biopsy tissue. The CCR score was calculated using a predefined linear combination of the CCP score and the Cancer of the Prostate Risk Assessment (CAPRA) score. The scores were evaluated for association with 10-year risk of metastatic disease following definitive therapy after adjusting for other clinical variables. RESULTS: The CCP score was strongly associated with 10-year risk of metastatic disease in multivariable analysis [Hazard Ratio per unit score = 2.21; 95% confidence interval (CI) 1.64, 2.98; p = 1.9 × 10-6] after adjusting for CAPRA, treatment type, and cohort. CCR was also highly prognostic (Hazard Ratio per unit score = 4.00; 95% CI 2.95, 5.42; p = 6.3 × 10-21). There was no evidence of interaction between CCP or CCR and cohort (p = 0.79 and p = 0.86, respectively) or treatment type (p = 0.55 and p = 0.78, respectively). Observed patient CCR-based predicted risks for metastatic disease by 10 years ranged from 0.1 to 99.4%, (IQR 0.7%, 4.6%). CONCLUSIONS: Both CCP and CCR scores provided independent prognostic information for predicting progression to metastatic disease after both surgery and radiation. These results further demonstrate their potential use as a risk stratification tool in patients with newly-diagnosed prostate cancer.

Comparison of the Prognostic Utility of the Cell Cycle Progression Score for Predicting Clinical Outcomes in African American and Non-African American Men with Localized Prostate Cancer.

BACKGROUND: Better prostate cancer risk stratification is necessary to inform medical management, especially for African American (AA) men, for whom outcomes are particularly uncertain. OBJECTIVE: To evaluate the utility of both a cell cycle progression (CCP) score and a clinical cell-cycle risk (CCR) score to predict clinical outcomes in a large cohort of men with prostate cancer highly enriched in an AA patient population. DESIGN, SETTING, AND PARTICIPANTS: Patients were diagnosed with clinically localized adenocarcinoma of the prostate and treated at The Ochsner Clinic (New Orleans, LA, USA) from January 2006 to December 2011. CCP scores were derived from archival formalin-fixed, paraffin-embedded biopsy tissue. CCR scores were calculated as the combination of molecular (CCP score) and clinical (Cancer of the Prostate Risk Assessment [CAPRA] score) components. INTERVENTION: Active treatment (radical prostatectomy, radiation therapy alone, or radiation and hormone therapy) or watchful waiting. OUTCOME MEASUREMENTS AND STATISTICAL ANALYSIS: The primary outcome was progression to metastatic disease. Association with outcomes was evaluated via Cox proportional hazards survival analysis and likelihood ratio tests. RESULTS AND LIMITATIONS: The final cohort included 767 men, of whom 281 (36.6%) were AA. After accounting for ancestry, treatment, and CAPRA in multivariable analysis, the CCP score remained a significant predictor of metastatic disease (hazard ratio [HR] 2.04; p<0.001), and there was no interaction with ancestry (p=0.20) or treatment (p=0.09). The CCR score was highly prognostic (HR 3.86; p<0.001), and as with the CCP score, there was no interaction with ancestry (p=0.24) or treatment (p=0.32). Limitations include the retrospective study design and the use of self-reported ancestry information. CONCLUSIONS: A CCR score provided significant prognostic information regardless of ancestry. The findings demonstrate that AA men in this study cohort appear to have similar prostate cancer outcomes to non-AA patients after accounting for all available molecular and clinicopathologic variables. PATIENT SUMMARY: In this study we evaluated the ability of a combined molecular and clinical score to predict the progression of localized prostate cancer. We found that the combined molecular and clinical score predicted progression to metastasis regardless of patient ancestry or treatment. This suggests that the combined molecular and clinical score may be a valuable tool for determining the risk of metastasis in men with newly diagnosed prostate cancer in order to make appropriate treatment decisions.

Identification of men with low-risk biopsy-confirmed prostate cancer as candidates for active surveillance.

BACKGROUND: A combined clinical cell-cycle risk (CCR) score that incorporates prognostic molecular and clinical information has been recently developed and validated to improve prostate cancer mortality (PCM) risk stratification over clinical features alone. As clinical features are currently used to select men for active surveillance (AS), we developed and validated a CCR score threshold to improve the identification of men with low-risk disease who are appropriate for AS. METHODS: The score threshold was selected based on the 90th percentile of CCR scores among men who might typically be considered for AS based on NCCN low/favorable-intermediate risk criteria (CCR = 0.8). The threshold was validated using 10-year PCM in an unselected, conservatively managed cohort and in the subset of the same cohort after excluding men with high-risk features. The clinical effect was evaluated in a contemporary clinical cohort. RESULTS: In the unselected validation cohort, men with CCR scores below the threshold had a predicted mean 10-year PCM of 2.7%, and the threshold significantly dichotomized low- and high-risk disease (P = 1.2 × 10-5). After excluding high-risk men from the validation cohort, men with CCR scores below the threshold had a predicted mean 10-year PCM of 2.3%, and the threshold significantly dichotomized low- and high-risk disease (P = 0.020). There were no prostate cancer-specific deaths in men with CCR scores below the threshold in either analysis. The proportion of men in the clinical testing cohort identified as candidates for AS was substantially higher using the threshold (68.8%) compared to clinicopathologic features alone (42.6%), while mean 10-year predicted PCM risks remained essentially identical (1.9% vs. 2.0%, respectively). CONCLUSIONS: The CCR score threshold appropriately dichotomized patients into low- and high-risk groups for 10-year PCM, and may enable more appropriate selection of patients for AS.

Analytical validation of a 12-gene molecular test for the prediction of distant recurrence in breast cancer.

AIM: To validate the analytical performance of a 12-gene molecular assay that predicts distant recurrence for early-stage ER+/HER2- invasive breast cancer as run within a central reference laboratory. MATERIALS & METHODS: Formalin-fixed paraffin-embedded breast resections were evaluated by quantitative reverse transcription polymerase chain reaction for the expression of eight target genes, three housekeeper genes and one control gene to assess for DNA contamination. RESULTS: The assay results were highly correlated with a validated reference laboratory. The assay had a broad linear range for input RNA, with similar amplicon efficiencies for target and housekeeper genes. The assay test was highly reproducible, with comparable inter- and intrabatch precision to the reference laboratory. CONCLUSION: These studies demonstrate that the 12-gene molecular assay is highly robust and accurate.

Cell cycle progression score predicts metastatic progression of clear cell renal cell carcinoma after resection.

BACKGROUND: The outcome of surgically resected, apparently localized, clear cell renal carcinoma (ccRCC) is uncertain. OBJECTIVE: To evaluate if cell cycle progression (CCP) gene expression can predict future metastasis. METHODS: Pathologic T2a-T3b tumors at University of Iowa were reviewed. Patients with known or suspected metastasis, lymph node involvement or who received neoadjuvant or adjuvant radiation, chemotherapy or immunotherapy were excluded. Case and control cohorts were defined as those who did or did not develop metastatic disease within 5 years. Measured levels of 31 cell cycle genes and 15 control genes from the tumor were calculated as a CCP score. Additionally, gene expression data for a separate ccRCC cohort was downloaded from The Cancer Genome Atlas (TCGA). RESULTS: Univariate analysis of 26 cases and 38 controls revealed that the CCP score predicted progression to metastasis (OR 2.65, p = 0.0091). In multivariate logistic regression modeling, CCP expression remained a significant independent predictor for progression (p = 0.026). The CCP score was also significantly associated with distant metastasis in the TCGA renal cancer cohort in both univariate (p = 1.0 × 10-9) and multivariate (p = 5.6 × 10-3) analysis. CONCLUSION: The CCP score has prognostic value in predicting metastatic progression after resection of organ-confined ccRCC.