SAN ANTONIO — Recurrent tumors carried distinct transcriptional programs compared with matched primary tumors, according to study results presented at the San Antonio Breast Cancer Symposium.
Nolan Priedigkeit, MD, of the department of pharmacology and chemical biology at the Women's Cancer Research Center of University of Pittsburgh Cancer Institute, suggested that longitudinal studies of gene expression in advanced tumors are lacking.
Researchers aimed to create a comprehensive characterization of genetic alterations in a cohort of 50 patient-matched pairs of primary and advanced ER-positive tumors using hybrid-capture RNA-sequencing. The analysis included 17 local tumors, along with various recurrence sites: 11 bone, 10 ovary, nine in the brain, and three gastrointestinal.
“Primary breast cancer does not equal recurrent or metastatic breast cancer,” Priedigkeit said. “Eventually tumors acquire therapeutic resistance. Molecular events may cause this difference in therapy response.”
Researchers analyzed 1,380 cancer-related genes for outlier expression fold-changes in matched recurrences vs. primary tumors.
The longest time to recurrence was 14.1 years, whereas the median time to recurrence was 3.4 years.
Based on observed pair-specific, outlier fold-changes, researchers assessed discrete, longitudinal transcriptional remodeling events for recurrence across all sites.
“The majority of genes don’t change,” Priedigkeit said. “We described extreme gains or losses at the edges of the curve as longitudinal transcriptional remodeling events.”
Researchers formed a hypothesis that recurrent breast cancers are molecularly distinct from primary tumors.
Results showed that 23 of 33 distant metastases retained PAM50 assignments of the matched primary. Shifts to HER-2 disease accounted for 12% of metastatic discordances, whereas shifts to luminal B comprised 15%.
Other findings showed “remarkably recurrent” gains and losses in terms of longitudinal transcriptional remodeling events at the genetic level in patients with advanced disease.
NCAM1 gains occurred among 42% of the cohort, whereas 40% showed gains in FGFR4, 36% in both IBSP and ROBO2, and 30% in SPP1. Conversely, longitudinal transcriptional modeling event losses occurred among 42% at RELN and 26% at ESR1.
The most significant enrichments were reported in NCAM1 (P < .001) among patients with distant disease (61% of n = 33), whereas these enrichments occurred among just one of 17 patients (6%) with locoregional disease.
“There are significant enrichments in metastatic tumors,” Priedigkeit said. He noted that challenges in studying metastatic cancers include heterogeneity, difficulty in associating genomic features with responses, use of nonindicated therapies, and obstacles in defining the natural history of tumor progression without patient-matched samples.
Other findings indicated an enriched longitudinal transcriptional remodeling event in brain metastases in RET (P = .003). This expression yielded outlier gains in 56% of ER-positive brain metastases.
RET inhibitor cabozantinib (Cabometyx, Exelixis) yielded antitumor activity in ex-vivo explant cultures among three patients with brain metastases who underwent resection and a brain-metastases patient-derived xenograft.
“Recurrent tumors carry distinct transcriptional programs vs. matched primary tumors,” Priedigkeit concluded. “Analyzing these outlier RNA expression deviations from primary tumors identifies acquired potentially targetable dependencies. Unique cancer subtypes may emerge after therapy selection. Highly recurrent longitudinal transcriptional remodeling events may challenge current emphasis on static, single time-point DNA-level biomarkers.” – by Rob Volansky
Priedigkeit N, et al. Abstract GS2-03. Presented at: San Antonio Breast Cancer Symposium; Dec. 5-9, 2017; San Antonio.
Disclosures: Priedigkeit reports no relevant financial disclosures. Please see the abstract for all other author’s relevant financial disclosures.