Abstract
Introduction: Currently, there are few markers available which are able to predict who is likely to respond to specific therapeutics in breast and lung cancer patients. This thesis describes candidate predictive markers, which would ultimately reduce cancer mortality as patients would receive therapy to which they respond from the start,
... read more
saving precious time. We used knowledge of well known molecular defects present in these cancer types to find predictive markers.
Methods: For breast cancer we used the knowledge that breast cancers with a mutation in BRCA1 or BRCA2, which also causes familial breast cancer, are hypersensitive to agents inducing double-strand breaks (DSBs) in the DNA. This is due to the fact that error-free repair of DNA DSBs depends on the homologous recombination DNA-repair pathway, of which BRCA1 and BRCA2 are part. BRCA1/2-mutated breast cancers have been shown to have characteristic DNA copy number gains and losses, resulting in specific genomic patterns, which can be visualized by array comparative genomic hybridization (aCGH). We evaluated whether these patterns would be able to select sporadic breast cancer patients benefitting from DSB-inducing therapy.
For lung cancer, we used the knowledge that patients with a mutation in the epidermal growth-factor receptor (EGFR) were shown to be responsive to EGFR-tyrosine-kinase inhibitors (EGFR-TKIs). We evaluated whether the level of soluble-EGFR and ligands of EGFR would predict benefit to EGFR-TKIs.
Results: For breast cancer, we tested genomic aCGH patterns of BRCA1-mutated (BRCA1-likeCGH), BRCA2-mutated (BRCA2-likeCGH), or either one (BRCA-likeCGH) in breast tumors as predictive markers in the HER2-negative, high-risk breast cancer population and found that: 1) one third of HER2-negative breast cancer patients had a BRCA-likeCGH tumor and half of these were hormone-receptor positive; 2) when tested in a randomized controlled trial, patients with BRCA-likeCGH breast cancers significantly benefited from high dose carboplatin-thiotepa-cyclophosphamide (CTC), a DSB-inducing regimen, compared to conventional anthracycline-based chemotherapy with regard to overall-survival, while patients without such patterns (non-BRCA-likeCGH) did not. This difference in treatment effect was significantly different (p-interaction), suggesting that BRCA-likeCGH status is a predictive marker; 3) In a BRCA1-associated mouse model low Xist RNA expression was significantly correlated with carboplatin hypersensitivity. Patients with low XIST RNA expression in their tumor showed improved survival after CTC therapy and more often displayed BRCA1-likeCGH genomic patterns.
In lung cancer high levels of soluble-EGFR and of amphiregulin (ARG) and low levels of transforming growth factor-alpha (TGFa) measured in serum were associated with better overall survival after EGFR-TKI treatment. With ARG and TGFa the difference in treatment effect was significantly different (p-interaction), implying these may be predictive markers.
Discussion: The studies described in this thesis advocate an approach for biomarker discovery which uses molecular aberrations in well-known oncogenic pathways as starting point, as these might be used in multiple cancer types. In the future, oncologists might treat cancer patients based on the evaluation of activation of oncogenic pathways, instead of per cancer type. These developments will largely depend on smart clinical trials implementing the search for new predictive biomarkers in its design, thereby ultimately increasing survival.
show less