Comprehensive molecular characterization of circulating tumor cells, extracellular vesicles and cell-free DNA as matched multi-parametric liquid biopsy for therapy management in metastatic breast cancer patients
Liquid biopsy analytes mirror tumoral heterogeneity and harbor great potential for minimal-invasive detection of biomarkers, useful for personalized therapeutic decisions and real-time monitoring. A spectrum of single liquid biopsy analytes has already been reported to present great potential for prognosis of metastatic breast cancer (MBC) patients. However, the utility of a combination of multiple analytes for highly sensitive therapy management remains elusive. The overall aim of this PhD thesis was the establishment of a multi-parametric workflow to isolate and analyze circulating tumor cell (CTC) mRNA, extracellular vesicle (EV) mRNA, CTC genomic DNA (gDNA) and cell-free DNA (cfDNA) from a minimized blood volume. The utility of this approach was compared to separated analysis of these single components for therapy management in 35 hormone receptor positive, HER2 negative MBC patients in the follow-up of the disease. Accordingly, the study was called ELIMA: Evaluation of multiple Liquid biopsy analytes In Metastatic breast cancer patients All from one blood sample.
mRNA profiles of CTCs and matched EVs were characterized by multimarker quantitative PCR. The comparison revealed substantial differences. While transcripts involved in the PI3K signal pathway were commonly overexpressed in CTCs, AURKA signals were highly prominent in EVs. mTOR signals in CTCs significantly correlated with response, whereas mTOR signals in EVs indicated therapy failure, reporting a contrary correlation of the same transcript in both analytes. Combinatorial analysis of ERBB2 and ERBB3 overexpression signals in CTCs and EVs was significantly associated with therapy failure, revealing the synergistic potential of both analytes.
Targeted deep sequencing of cfDNA using at least 4 ml plasma for sufficient cfDNA input for library preparation revealed high inter-patient variability of cfDNA concentration. The integration of unique molecular indicies was demonstrated to guarantee specificity of called variants. Comparison of cfDNA variants isolated from whole blood and matched CTC-depleted blood presented no significant differences, enabling a reliable combined CTC and cfDNA analysis from only 10 ml of blood. cfDNA variant analysis in MBC patients displayed a high prevalence of MUC16 variants. 25% of all detected variants were known to be pathogenic and likely pathogenic, mostly located in AR, BRCA1, ESR1 and PIK3CA. The presence of the specific ESR1 Y539S and PIK3CA E545K variants was significantly correlated with decreased time between first diagnosis of metastases and death. The dynamics of ESR1 and PIK3CA variant allele frequencies across treatment were reported to harbor great potential as a sensitive monitoring approach.
Comprehensive experiments were performed to establish a protocol for CTC gDNA isolation, also enabling parallel CTC mRNA isolation. We optimized binding and washing conditions for gDNA purification by QIAamp MinElute spin columns, we excluded a whole genome amplification step and modified the library preparation protocol, facilitating variant calling of CTC gDNA by targeted deep sequencing. Despite a challenging low number of unique CTC gDNA fragments sequenced, a recovery of 57% of all cfDNA variants was observed in the CTC gDNA. Of all detected variants in both, the cfDNA and CTC gDNA, 16% were shared. Consequently, 84% unique variants revealed the complementary nature of cfDNA and CTC gDNA.
The comparison of CTC mRNA, EV mRNA, CTC gDNA and cfDNA in matched MBC samples exhibited advantages of using one of the particular analytes depending on specific clinical questions. Although a higher sensitivity of overexpression and variant analysis in the CTC fraction compared to EVs and cfDNA in general, a similar percentage (range 50%-64%) of patients was identified with actionable markers in all four single analytes. Finally, the integration of information drawn from the multi-parametric protocol uncovered that only this extensive approach enabled the identification of actionable markers in 93% of all patients, confirming the synergistic value of the examined liquid biopsy analytes. In the ELIMA study, the successfully established workflow for multi-parametric liquid biopsy evaluation (Figure 31) facilitated the insight about a comprehensive picture of the transcriptomic and genomic complexity in each individual patient, feasible for sensitive, personalized therapy management in MBC in the future.
mRNA profiles of CTCs and matched EVs were characterized by multimarker quantitative PCR. The comparison revealed substantial differences. While transcripts involved in the PI3K signal pathway were commonly overexpressed in CTCs, AURKA signals were highly prominent in EVs. mTOR signals in CTCs significantly correlated with response, whereas mTOR signals in EVs indicated therapy failure, reporting a contrary correlation of the same transcript in both analytes. Combinatorial analysis of ERBB2 and ERBB3 overexpression signals in CTCs and EVs was significantly associated with therapy failure, revealing the synergistic potential of both analytes.
Targeted deep sequencing of cfDNA using at least 4 ml plasma for sufficient cfDNA input for library preparation revealed high inter-patient variability of cfDNA concentration. The integration of unique molecular indicies was demonstrated to guarantee specificity of called variants. Comparison of cfDNA variants isolated from whole blood and matched CTC-depleted blood presented no significant differences, enabling a reliable combined CTC and cfDNA analysis from only 10 ml of blood. cfDNA variant analysis in MBC patients displayed a high prevalence of MUC16 variants. 25% of all detected variants were known to be pathogenic and likely pathogenic, mostly located in AR, BRCA1, ESR1 and PIK3CA. The presence of the specific ESR1 Y539S and PIK3CA E545K variants was significantly correlated with decreased time between first diagnosis of metastases and death. The dynamics of ESR1 and PIK3CA variant allele frequencies across treatment were reported to harbor great potential as a sensitive monitoring approach.
Comprehensive experiments were performed to establish a protocol for CTC gDNA isolation, also enabling parallel CTC mRNA isolation. We optimized binding and washing conditions for gDNA purification by QIAamp MinElute spin columns, we excluded a whole genome amplification step and modified the library preparation protocol, facilitating variant calling of CTC gDNA by targeted deep sequencing. Despite a challenging low number of unique CTC gDNA fragments sequenced, a recovery of 57% of all cfDNA variants was observed in the CTC gDNA. Of all detected variants in both, the cfDNA and CTC gDNA, 16% were shared. Consequently, 84% unique variants revealed the complementary nature of cfDNA and CTC gDNA.
The comparison of CTC mRNA, EV mRNA, CTC gDNA and cfDNA in matched MBC samples exhibited advantages of using one of the particular analytes depending on specific clinical questions. Although a higher sensitivity of overexpression and variant analysis in the CTC fraction compared to EVs and cfDNA in general, a similar percentage (range 50%-64%) of patients was identified with actionable markers in all four single analytes. Finally, the integration of information drawn from the multi-parametric protocol uncovered that only this extensive approach enabled the identification of actionable markers in 93% of all patients, confirming the synergistic value of the examined liquid biopsy analytes. In the ELIMA study, the successfully established workflow for multi-parametric liquid biopsy evaluation (Figure 31) facilitated the insight about a comprehensive picture of the transcriptomic and genomic complexity in each individual patient, feasible for sensitive, personalized therapy management in MBC in the future.