Molecular Analysis of Breast Cancer Transcriptomes, Genomes, and Circulating Tumor DNA

Detta är en avhandling från Division of Oncology and Pathology

Sammanfattning: Breast cancer is a very heterogeneous disease in terms of clinical characteristics, genetic aberrations and prognosis. In Paper I, we focused on the CD44 molecule that often is aberrantly expressed in breast cancer and is widely used as a marker for cancer stem cells. Several isoforms of the CD44 molecule were analyzed at the transcriptome level across breast tumors and the expression of individual isoforms was correlated to molecular subtypes, protein expression of clinical markers, and cancer stem cell (CSC) phenotypes in breast tumors and cell lines. The CD44S isoform was associated with expression of the CSC marker ALDH1 and the CSC phenotype CD44+/CD24- was correlated to alternatively spliced isoforms in tumors. The isoforms were differentially expressed in molecular subtypes and HER2 and EGFR positive tumors were associated to CD44S and CD44v8-10, respectively. In Paper II, by using targeted genomic re-sequencing we screened for somatic mutations in 1237 genes in a panel of basal-like breast cancer cell lines, both in coding and surrounding non-coding regions. In total, 658 high confidence SNVs and indels were detected and 315 of these were novel (not in COSMIC). A selection of the variants were validated with Sanger sequencing and, 123 of 130 high confidence variants were confirmed including 111 novel variants. The mutation frequency was higher in coding (CDS) compared to non-coding (non- CDS) regions and in particular G or C base replacements were higher in the CDS compared to non-CDS. The SNVs within the context of T[C]A/T[G]A and T[C]T/A[G]A were significantly more common in the CDS than in the non-CDS regions. Re-sequenced data was used to derive copy number estimations, which correlated well to SNP array data. In Paper III, the potential in using tumor-specific rearrangements present in circulating tumor DNA (ctDNA) to detect occult metastatic breast cancer was evaluated. In total, 14 eventual metastatic (EM) patients and 6 long-term disease free (DF) patients were investigated. We used whole-genome sequencing on the primary tumors to derive patient-specific rearrangements that were confirmed by PCR. Circulating tumor DNA levels across multiple plasma samples during the clinical course were analyzed by quantitative droplet digital PCR. Accurate post-surgical discrimination of EM patients (93%) from DM (100%) was achieved by ctDNA monitoring. The average lead-time to clinical detection of metastatic disease was 11 months (range 0-37 months). Moreover, the ctDNA level was a quantitative predictor for both recurrence (P=0.02) and death (P=0.04). We demonstrated that monitoring of ctDNA can be used for early detection of metastatic breast cancer and is a potential tool for optimization of adjuvant therapy and should be evaluated further in clinical studies.