Multi-cohort modeling strategies for scalable globally accessible prostate cancer risk tools

BACKGROUND

Online clinical risk prediction tools built on data from multiple cohorts are increasingly being utilized for contemporary doctor-patient decision-making and validation. This report outlines a comprehensive data science strategy for building such tools with application to the Prostate Biopsy Collaborative Group prostate cancer risk prediction tool.

METHODS

We created models for high-grade prostate cancer risk using six established risk factors. The data comprised 8492 prostate biopsies collected from ten institutions, 2 in Europe and 8 across North America. We calculated area under the receiver operating characteristic curve (AUC) for discrimination, the Hosmer-Lemeshow test statistic (HLS) for calibration and the clinical net benefit at risk threshold 15%. We implemented several internal cross-validation schemes to assess the influence of modeling method and individual cohort on validation performance.

RESULTS

High-grade disease prevalence ranged from 18% in Zurich (1863 biopsies) to 39% in UT Health San Antonio (899 biopsies). Visualization revealed outliers in terms of risk factors, including San Juan VA (51% abnormal digital rectal exam), Durham VA (63% African American), and Zurich (2.8% family history). Exclusion of any cohort did not significantly affect the AUC or HLS, nor did the choice of prediction model (pooled, random-effects, meta-analysis). Excluding the lowest-prevalence Zurich cohort from training sets did not statistically significantly change the validation metrics for any of the individual cohorts, except for Sunnybrook, where the effect on the AUC was minimal. Therefore the final multivariable logistic model was built by pooling the data from all cohorts using logistic regression. Higher prostate-specific antigen and age, abnormal digital rectal exam, African ancestry and a family history of prostate cancer increased risk of high-grade prostate cancer, while a history of a prior negative prostate biopsy decreased risk (all p-values < 0.004).

CONCLUSIONS

We have outlined a multi-cohort model-building internal validation strategy for developing globally accessible and scalable risk prediction tools.

Estrogen promotes estrogen receptor negative BRCA1-deficient tumor initiation and progression

BACKGROUND

Estrogen promotes breast cancer development and progression mainly through estrogen receptor (ER). However, blockage of estrogen production or action prevents development of and suppresses progression of ER-negative breast cancers. How estrogen promotes ER-negative breast cancer development and progression is poorly understood. We previously discovered that deletion of cell cycle inhibitors p16Ink4a (p16) or p18Ink4c (p18) is required for development of Brca1-deficient basal-like mammary tumors, and that mice lacking p18 develop luminal-type mammary tumors.

METHODS

A genetic model system with three mouse strains, one that develops ER-positive mammary tumors (p18 single deletion) and the others that develop ER-negative tumors (p16;Brca1 and p18;Brca1 compound deletion), human BRCA1 mutant breast cancer patient-derived xenografts, and human BRCA1-deficient and BRCA1-proficient breast cancer cells were used to determine the role of estrogen in activating epithelial-mesenchymal transition (EMT), stimulating cell proliferation, and promoting ER-negative mammary tumor initiation and metastasis.

RESULTS

Estrogen stimulated the proliferation and tumor-initiating potential of both ER-positive Brca1-proficient and ER-negative Brca1-deficient tumor cells. Estrogen activated EMT in a subset of Brca1-deficient mammary tumor cells that maintained epithelial features, and enhanced the number of cancer stem cells, promoting tumor progression and metastasis. Estrogen activated EMT independent of ER in Brca1-deficient, but not Brca1-proficient, tumor cells. Estrogen activated the AKT pathway in BRCA1-deficient tumor cells independent of ER, and pharmaceutical inhibition of AKT activity suppressed EMT and cell proliferation preventing BRCA1 deficient tumor progression.

CONCLUSIONS

This study reveals for the first time that estrogen promotes BRCA1-deficient tumor initiation and progression by stimulation of cell proliferation and activation of EMT, which are dependent on AKT activation and independent of ER.

A phase II study of the dual mTOR inhibitor MLN0128 in patients with metastatic castration resistant prostate cancer

Background MLN0128 is a first-in-class, dual mTOR inhibitor with potential to outperform standard rapalogs through inhibition of TORC1 and TORC2. This phase II study was designed to assess antitumor activity of MLN0128 in metastatic castration-resistant prostate cancer (mCRPC). Methods Eligible patients had mCRPC previously treated with abiraterone acetate and/or enzalutamide. Five patients started MLN0128 at 5 mg once daily, subsequently dose reduced to 4 mg because of toxicity. Four subsequent patients started MLN0128 at 4 mg daily. Primary endpoint was progression-free survival at 6 months. Results Nine patients were enrolled and median time on treatment was 11 weeks (range: 3-30). Best response was stable disease. All patients had a rise in PSA on treatment, with a median 159% increase from baseline (range: 12-620%). Median baseline circulating tumor cell count was 1 cell/mL (range: 0-40); none had a decrease in cell count posttreatment. Grade ≤ 2 adverse events included fatigue, anorexia, and rash. The most common serious adverse events were grade 3 dyspnea and maculopapular rash. Eight patients discontinued treatment early because of radiographic progression (n = 1), grade 3 toxicity (n = 5), or investigator discretion (n = 2). Four patients had immediate PSA decline following drug discontinuation, suggesting MLN0128 could cause compensatory increase of androgen receptor (AR) activity. Correlative studies of pretreatment and posttreatment biopsy specimens revealed limited inhibition of AKT phosphorylation, 4EBP1 phosphorylation, and eIF4E activity. Conclusions Clinical efficacy of MLN0128 in mCRPC was limited likely due to dose reductions secondary to toxicity, PSA kinetics suggesting AR activation resulting from mTOR inhibition, and poor inhibition of mTOR signaling targets.

A mutation in the viral sensor 2'-5'-oligoadenylate synthetase 2 causes failure of lactation

We identified a non-synonymous mutation in Oas2 (I405N), a sensor of viral double-stranded RNA, from an ENU-mutagenesis screen designed to discover new genes involved in mammary development. The mutation caused post-partum failure of lactation in healthy mice with otherwise normally developed mammary glands, characterized by greatly reduced milk protein synthesis coupled with epithelial cell death, inhibition of proliferation and a robust interferon response. Expression of mutant but not wild type Oas2 in cultured HC-11 or T47D mammary cells recapitulated the phenotypic and transcriptional effects observed in the mouse. The mutation activates the OAS2 pathway, demonstrated by a 34-fold increase in RNase L activity, and its effects were dependent on expression of RNase L and IRF7, proximal and distal pathway members. This is the first report of a viral recognition pathway regulating lactation.

Using ENU-mutagenesis in mice we discovered a pedigree with lactation failure. Mammary development through puberty and pregnancy appeared normal in mutant animals, but the activation of lactation failed in the immediate post partum period and no milk reached the pups. Failure of lactation was accompanied by greatly diminished milk protein synthesis, decreased epithelial cell proliferation, increased epithelial cell death and a robust interferon response. A non-synonymous mutation in Oas2 (I405N) in the viral sensor Oas2 was found and expression of mutant Oas2 in mammary cells recapitulated these phenotypes. RNase L, the most proximal effector of OAS2 action, was activated in the mammary glands of mutant mice and in mammary cells expressing mutant Oas2. Knockdown of RNase L, or the distal pathway member IRF7, prevented these effects, indicating that the mutation in OAS2 caused activation of the viral signaling pathway. These results show that viral detection in the mammary gland can prevent lactation.