Genitourinary Cancer

SS 28 - GU 3 - New Insights Into Treatment Intensification Strategies for Prostate Cancer

200 - Development and Validation of Genomic Tools to Predict Extraprostatic Extension of Prostate Cancer, Opportunities for Personalizing Treatment

Tuesday, October 23
2:55 PM - 3:05 PM
Location: Room 007 A/B

Development and Validation of Genomic Tools to Predict Extraprostatic Extension of Prostate Cancer, Opportunities for Personalizing Treatment
W. A. Hall1, S. Liu2, N. Fishbane2, M. J. Xu3, E. Davicioni2, B. A. Mahal4, R. B. Den5, R. T. Dess6, W. C. Jackson6, A. C. Wong7, E. M. Schaeffer8, R. J. Karnes9, P. Carroll10, P. L. Nguyen11, C. A. F. Lawton12, D. E. Spratt6, and F. Y. Feng7; 1Medical College of Wisconsin and Clement J Zablocki VA Medical Center, Milwaukee, WI, 2GenomeDx Biosciences, Vancouver, BC, Canada, 3University of California San Francisco, Department of Radiation Oncology, San Francisco, CA, 4Harvard Radiation Oncology Program, Harvard Medical School, Boston, MA, 5Sidney Kimmel Medical College at Thomas Jefferson University, Sidney Kimmel Cancer Center, Philadelphia, PA, 6Department of Radiation Oncology, University of Michigan, Ann Arbor, MI, 7Department of Radiation Oncology, University of California San Francisco, San Francisco, CA, 8Northwestern University, Evanston, IL, 9Department of Urology, Mayo Clinic, Rochester, MN, 10University of California San Francisco, Department of Urology, San Francisco, CA, 11Department of Radiation Oncology, Dana-Farber Cancer Institute and Brigham and Women's Hospital, Harvard Medical School, Boston, MA, 12Medical College of Wisconsin Department of Radiation Oncology, Milwaukee, WI

Purpose/Objective(s): Pretreatment estimates of seminal vesicle invasion (SVI) and extra-capsular extension (ECE) can be used to guide treatment recommendations for prostate cancer (PCa). Current predictive nomograms for SVI and ECE, such as the Memorial Sloan Kettering (MSK) and Partin tables, use Gleason score, PSA, and T-stage. We sought to improve the ability of these nomograms to predict ECE or SVI via the incorporation of novel genomic signatures.

Materials/Methods: 15,609 prostatectomy (RP) samples were divided into three sets: a training set of PCa samples used for model development (6,938), a test set model evaluation (n=3,469), along with an independent validation sets from RP (n=5,202). In addition a separate cohort of biopsy samples (n=698) was used for validation. The RP samples were from men with adverse pathology after surgery and the biopsy cohort consisted of men diagnosed with cT1-cT2 disease, treated with RP and had complete pathological records available. Transcriptome-wide analysis (HuEx 1.0 ST array) was performed on all PCa samples and retrieved from the de-identified GRID registry (NCT02609269). Two ridge regression models were developed from the training and test sets, one to predict SVI (581 gene features) and a second to predict ECE (533 gene features). The model was evaluated using AUC analysis including optimism adjustment when combined with clinical nomograms. Multivariate logistic regression was performed to assess the predictive effect of the genomic (ECE/SVI) classifier in the presence of established nomograms.

Results: The biopsy validation set consisted primarily of NCCN low and favorable intermediate risk disease (72%) at diagnosis and after surgery 28% and 7% had ECE or SVI. The genomic ECE and SVI models had AUC of 0.68 (95% CI: 0.64-0.73) and 0.78 (0.71-0.85). When comparing AUC values for ECE and SVI both the Partin and MSK nomograms showed improvement in their AUC values with incorporation of genomic signatures. The AUC of Partin + Genomic ECE was 0.70 (0.66-0.75), higher than Partin alone (AUC 0.62) and Partin + Genomic SVI was 0.86 (0.79-0.92), higher than Partin alone (AUC 0.81). Similar results were observed with the MSK nomogram. Multivariate logistic regression adjusting for the clinical nomograms showed the genomic ECE and SVI models had significant odds ratios (OR) for predicting the endpoints at RP. For example, adjusting for Partin the genomic ECE and SVI models had OR 1.60 (p<0.001) and 2.48 (p<0.001) for each 10% increase in score.

Conclusion: Using genomic data from a large prospective registry, we have developed and validated novel expression signatures to predict the presence of either ECE or SVI prior to treatment. These genomic tools add significantly to existing clinical predictive nomograms, and potentially have important implications in predicting the need for post-operative radiation therapy in patients treated with upfront surgery. Such data may also improve treatment volumes in patients managed with definitive radiation therapy.

Author Disclosure: W.A. Hall: None. S. Liu: Stock; GenomeDx Biosciences Inc. N. Fishbane: Stock; GenomeDx Biosciences Inc. M.J. Xu: None. E. Davicioni: Stock; GenomeDx Biosciences Inc. President; GenomeDx Biosciences Inc. B.A. Mahal: None. R.B. Den: Research Grant; GenomeDx. Speaker's Bureau; Bayer. Advisory Board; GenomeDx, Bayer. W.C. Jackson: None. E.M. Schaeffer: None. R.J. Karnes: None. P. Carroll: Research Grant; Genomic Health International, Myriad. Honoraria; Genomic Health International, Janssen, Takeda. Advisory Board; Genomic Health International. Board of Directors; National Comprehensive Cancer Network. P.L. Nguyen: Honoraria; Bayer. Consultant; Nanobiotix, Infinity Pharmaceuticals, GI Windows, Astellas, Augmenix. Advisory Board; Ferring, Medivation, Genome DX, Dendreon. Stock Options; Augmenix. Program Committee; Genitourinary Cancers Symposium. F.Y. Feng: Research Grant; GenomeDx. Advisory Board; GenomeDx, Dendreon, Sanofi. Travel Expenses; GenomeDx. Liaison, GU Translational Research Program; Radiation Therapy Oncology Group. President and Founder; PFS Genomics.

William Hall, MD

Medical College of Wisconsin

Disclosure:
Employment
Medical College of Wisconsin: Assistant Professor: Employee, Assistant Professor of Radiation Oncology: Employee, Assistant Professor of Radiation Oncologyr: Employee

Biography:
I am a board certified radiation oncologist with a passion for research in gastrointestinal malignancies, and radiological sciences. Thus far, in the early portion of my career, I have devoted my efforts to understanding the integration of advanced imaging modalities, including MRI, into radiation treatment planning. Moreover, I have published several peer-reviewed articles devoted to the optimal use of radiation therapy for gastrointestinal malignancies and prostate cancer. Such articles have specifically related to radiation therapy technological advances, MRI incorporation into radiation planning, and the optimal radiation therapy dosing strategy. My future research goals are focused on developing and improving our ability to identify patients that may not require surgery for rectal cancer. In addition, I hope to develop MRI strategies that would use novel and unique MRI sequences to enable precise delineation of the most malignant portions of a tumor. My aspiration is to work closely with collaborators to refine our ability to accurately identify malignant tissue and direct intensification of local radiation therapy. I see the next decade of research to be an exciting and transformative time in the management of several different malignancies and I am looking forward to making whatever contributions I can.

Presentation(s):

Send Email for William Hall


Assets

200 - Development and Validation of Genomic Tools to Predict Extraprostatic Extension of Prostate Cancer, Opportunities for Personalizing Treatment



Attendees who have favorited this

Please enter your access key

The asset you are trying to access is locked. Please enter your access key to unlock.

Send Email for Development and Validation of Genomic Tools to Predict Extraprostatic Extension of Prostate Cancer, Opportunities for Personalizing Treatment