Leslie R. Euceda1, Tonje H. Haukaas1,2, Guro F. Giskeødegård1, Riyas Vettukattil1, Geert Postma3, Laxmi Silwal-Pandit2,4, Jasper Engel5, Lutgarde M.C. Buydens3, Anne-Lise Børresen-Dale2,4, Olav Engebraaten6, and Tone F. Bathen1,2
1Department of Circulation and Medical Imaging, The Norwegian University of Science and Technology, Trondheim, Norway, 2K.G. Jebsen Center for Breast Cancer Research, Institute of Clinical Medicine, University of Oslo, Oslo, Norway, 3Institute for Molecules and Materials, Radboud University Nijmegen, Nijmegen, Netherlands, 4Department of Genetics, Institute for Cancer Research, Oslo University Hospital, The Norwegian Radium Hospital, Oslo, Norway, 5NERC Biomolecular Analysis Facility Metabolomics Node (NBAF-B), School of Biosciences, University of Birmingham, Birmingham, United Kingdom, 6Department of Oncology, Department of Tumor Biology, Oslo University Hospital, Oslo, Norway
Synopsis
This study used HR MAS magnetic resonance based metabolic profiles from breast tumor tissue to explore the metabolic changes occurring as an effect of overall neoadjuvant therapy, discriminate therapy responders from nonresponders, and determine metabolic differences between patients receiving or not receiving the antiangiogenic drug bevacizumab. Changes as an effect of chemotherapy were detected and responders were successfully discriminated from nonresponders after treatment, showing potential for assessment of patient benefit to treatment and the understanding of underlying mechanisms affecting response. Although metabolic differences based on bevacizumab administration were not prominent, glutathione was identified to be possibly affected by the drug.Purpose
Metabolomics investigates biochemical processes
influenced by environmental factors at a level closer to the phenotype compared
to transcriptomics and proteomics, potentially providing complimentary insight
into treatment outcome. This study aimed to use magnetic resonance (MR) based
metabolic profiles from breast tumor tissue to explore the metabolic changes
occurring as an effect of overall neoadjuvant therapy, discriminate therapy
responders from nonresponders, and determine metabolic effects of the
antiangiogenic drug bevacizumab.Methods
The metabolic profiles of 122 breast cancer patients (n=270 tissue samples) were determined by high resolution (HR) magic angle spinning (MAS) MR spectroscopy. All patients received FEC (5-fluorouracil-epirubicin-cyclophosphamide) and taxanes as chemotherapeutic neoadjuvant treatment, while they were randomized to receive bevacizumab or not (controls). Biopsies were sampled prior to treatment (TP1), after 12 weeks (TP2), and at surgery (TP3). Samples were classified into gene expression subtypes using the PAM50 algorithm1. Multivariate strategies were used to analyze the metabolic profiles. Levels of 16 metabolites were calculated by integration and analyzed by univariate linear mixed models (LMM)2.Results
Principal component analysis (PCA) showed clear changes in metabolic profiles (Figure 1) with glucose and lipids increasing as an effect of chemotherapy. Partial least squared-discriminant analysis (PLS-DA) revealed metabolic differences between responders and nonresponders at TP3, but not TP1 or TP2, with an accuracy of 77% (p <0.001). Metabolic profiles at TP1 and TP3 discriminated patients with ≥90% tumor reduction (TP3size/TP1size) from those with ≤10% reduction with an accuracy of 76% (p=0.001) and 75% (p=0.002), respectively. A metabolic switch before and after treatment was observed between these groups, with lower glucose (Glc) and higher lactate (Lac) in the ≥90% reduction group at TP1 (Figure 2). Bevacizumab-receiving patients and controls, and bevacizumab-receiving responders and control responders could not be discriminated at any time point. Multilevel-PLSDA (paired multivariate analysis) showed significant separation (p<0.05) of earlier time points from later time points in all cases, confirming a metabolic effect due to treatment as observed in PCA. However, no metabolic differences could be detected between responders and nonresponders nor bevacizumab-receiving patients and controls with the paired analyses. Unpaired LMM revealed significant differences in 11/16 metabolites for the factor time and 8/16 for response. A significant interaction between time and bevacizumab for glutathione revealed higher levels of this antioxidant in controls than in bevacizumab-treated patients after treatment (Figure 3).Discussion
The increase of glucose with treatment progression points to a decline in glucose consumption, which is characteristically rapid in cancer cells3. The increase of lipids towards TP3 suggests a transition to normal breast adipose tissue, which is supported by samples from TP3 and of the characteristically adipose-enriched normal-like subtype4 corresponding well in the PCA scores plot (Figure 1). Despite metabolic profiles not being able to predict response prior to treatment, a significant metabolic difference in responders compared to nonresponders was detected after neoadjuvant chemotherapy. Furthermore, a metabolic switch at TP1 and TP3 between ≥90% and ≤10% reduction groups was observed, indicating a higher Warburg effect3 in the former group before treatment, which is subsequently reversed by the end of treatment. Compounds associated with a more malignant phenotype including the choline-containing metabolites and succinate (Succ) were elevated in the ≥90% reduction group before treatment, while the opposite was observed at TP3. This suggests that patients with a seemingly more malignant metabolic profile are more prone to benefit from treatment in terms of tumor reduction. Changes due to bevacizumab administration were not as straightforward as the rest of our findings. The substantial metabolic effect induced by chemotherapy may mask the effects of bevacizumab. However, glutathione was decreased in patients receiving bevacizumab. Bevacizumab may thus play a role in redox destabilization.
Conclusions
MR spectroscopy metabolic profiles reflected changes as an effect of chemotherapy and successfully discriminated responders from nonresponders after treatment, showing potential for assessment of patient benefit to treatment and the understanding of underlying mechanisms affecting response. Although metabolic differences based on bevacizumab administration were not prominent, the study indicates that glutathione is affected by the antiangiogenic drug.Acknowledgements
This work is supported by grants from the Research Council of Norway (FRIMEDBIO221879).References
1] Parker JS, et al. Supervised Risk Predictor of Breast Cancer Based on Intrinsic Subtypes, J Clin Oncol, 2009; 27(8):1160-1167. 2] Pinheiro JC, et al. Mixed-Effects Models in S and S-PLUS. Statistics and Computing. New York, NY, USA: Springer New York; 2000. p. 3-56 2. 3] Vander Heiden MG, et al. Understanding the Warburg Effect: The Metabolic Requirements of Cell Proliferation . Science. 2009;324(5930):1029-33. 3. 4] Perou CM, et al. Molecular portraits of human breast tumours. Nature. 2000;406(6797):747-52.