Main Session
Sep 29
PQA 06 - Genitourinary Cancer, Gynecological Cancer, and Health Care Access and Engagement

3343 - Mechanistic Insights into Radiopharmaceutical Therapy Resistance through Bayesian Analysis of Circulating Tumor Cell Transcriptomes

02:15pm - 03:30pm ET
Poster Hall - Exhibit Hall A
Screen: 15
POSTER

Presenter(s)

Keisuke Otani, MD, PhD - Massachusetts General Hospital, Charlestown, Massachusetts

K. Otani1, H. Hirata1,2, Y. Otani1, B. Owusu-Amo1, S. Qin1, M. Lemieux1, D. J. Rodden1, R. J. Lee3, P. J. Saylor4, and D. T. Miyamoto1; 1Department of Radiation Oncology, Massachusetts General Hospital, Harvard Medical School, Boston, MA, 2Department of Radiation Oncology, National Cancer Center Hospital East, Kashiwa, Japan, 3Massachusetts General Hospital, Boston, MA, 4Department of Medicine, Mass General Brigham / Massachusetts General Hospital, Boston, MA

Purpose/Objective(s):

The radiopharmaceutical radium-223 (Ra 223) improves survival in metastatic prostate cancer, but not all patients respond and resistance mechanisms are poorly understood. Circulating tumor cells (CTCs) offer a means to non-invasively probe cancer biological state and predict therapeutic response, but their transcriptomic analysis is complicated by leukocyte contamination. Here we employed microfluidic enrichment of CTCs, whole transcriptome RNA-seq, and computational deconvolution using Bayesian cell proportion reconstruction to identify biological pathways associated with resistance to Ra 223 in metastatic prostate cancer.

Materials/Methods:

Twenty-two men with prostate cancer with bone-predominant metastases treated with Ra 223 were enrolled in this prospective exploratory biomarker study. CTCs were isolated from pre-treatment blood using a negative selection microfluidic device. RNA was purified and prepared for bulk RNA-seq. Bayesian cell proportion reconstruction inferred using statistical marginalization (BayesPrism) was used to deconvolute CTC and leukocyte signals from bulk RNA-seq data. As reference for deconvolution, single cell RNA-seq data for prostate cancer, leukocytes, and platelets were obtained from public databases. The method was validated using RNA-seq data from prostate cancer cell lines (LNCaP) spiked into healthy donor blood. Six patients with survival less than 1 year were defined as extreme non-responders, while the 6 longest surviving patients were defined as extreme responders.

Results:

BayesPrism estimates of tumor-derived cell proportions showed improved correlation with spiked LNCaP cell input numbers compared to older deconvolution methods (Spearman’s rho = 0.8, R2 = 0.67). Deconvoluted expression of individual cancer genes (AR, STEAP2, TMPRSS2) showed strong correlation with the number of spiked LNCaP cells (rho = 0.8 - 1, R2 = 0.974-0.999). Differential gene expression analysis of CTCs at baseline in patients treated with Ra 223 identified 69 upregulated genes in extreme non-responders compared to responders. Biological process analyses using Gene Ontology and Gene Set Enrichment Analysis (GSEA) revealed enrichment of multiple mitochondria-related terms. Single sample GSEA for hallmark oxidative phosphorylation gene set (OXPHOS) in 22 patients showed significantly worse survival after Ra 223 in patients with high OXPHOS CTC signatures (p = 0.0094). Similar analysis for DNA repair signatures showed a trend towards worse survival in patients with CTCs with signatures of high DNA repair activity (p = 0.097).

Conclusion:

Activated OXPHOS was significantly correlated with poor survival after Ra 223 in metastatic prostate cancer. These findings support the potential of CTC analyses to identify radiopharmaceutical resistance mechanisms, although further validation is needed.