Path analysis of intra-metastatic hypoxia in breast cancer

Abstract

AbstractHypoxia (low O2) signals into the nucleus of cancer cells through hypoxia-inducible factor (HIF)-1α-dependent transcription triggering proliferative, metabolic and vascular adaptations linked to therapy resistance and mortality due to overt metastasis. In contrast with the wealth of molecular data on primary intra-tumoral hypoxia, there is a dearth of statistical modelling studies addressing the mechanisms of intra-metastatic hypoxia. In this study, we used path analysis to model intra-metastatic hypoxia (Hx), HIF-1α expression and microvascular area (MVA) as functions of metastatic cross-sectional area (MCSA) in an advanced mouse breast cancer model; in this context, we tested the effect of conventional, maximum-tolerated dose (MTD) or low-dose metronomic (LDM) chemotherapy. Iterative analysis of 34 non-isomorphic paths yielded four well-fitting, configuration-invariant models [χ2(6,171)≥ 6.12;P≥ 0.328; CFI ≥ 0.998; RMSEA ≤ 0.07]. All four models contained HIF-1α as a mediating variable within the MCSA↔Hx↔MVA path, as well as significant Hx↔MVA interactions. LDM disrupted the HIF-1α↔MCSA→Hx and HIF-1α↔MVA paths; furthermore, all LDM and MTD combinations impaired Hx↔HIF-1α. These results confirmed well-established hypoxic interactions, whilst uncovering possible differential effects of chemotherapeutic modalities upon metastatic size, hypoxia, HIF-1α and vascularisation. Our data indicate that MVA can act as a downstream readout, rather than an adaptive angiogenic mechanism alleviating intra-metastatic hypoxia. Moreover, well-fitting path models locate HIF-1α activity either upstream or downstream of hypoxia, thereby allowing us to posit the existence of bi-directional feedback loops driving vascularisation and growth in metastatic tumours, of relevance for targeted therapies.