Search

Purpose: Apoptosis is essential for chemotherapy responses. In this discovery and validation study, we evaluated the suitability of a mathematical model of apoptosis execution (APOPTO-CELL) as a stand-alone signature and as a constituent of further refined prognostic stratification tools.Experimental Design: Apoptosis competency of primary tumor samples from patients with stage III colorectal cancer (n = 120) was calculated by APOPTO-CELL from measured protein concentrations of Procaspase-3, Procaspase-9, SMAC, and XIAP. An enriched APOPTO-CELL signature (APOPTO-CELL-PC3) was synthesized to capture apoptosome-independent effects of Caspase-3. Furthermore, a machine learning Random Forest approach was applied to APOPTO-CELL-PC3 and available molecular and clinicopathologic data to identify a further enhanced signature. Association of the signature with prognosis was evaluated in an independent colon adenocarcinoma cohort (TCGA COAD, n = 136).Results: We identified 3 prognostic biomarkers (P = 0.04, P = 0.006, and P = 0.0004 for APOPTO-CELL, APOPTO-CELL-PC3, and Random Forest signatures, respectively) with increasing stratification accuracy for patients with stage III colorectal cancer.The APOPTO-CELL-PC3 signature ranked highest among all features. The prognostic value of the signatures was independently validated in stage III TCGA COAD patients (P = 0.01, P = 0.04, and P = 0.02 for APOPTO-CELL, APOPTO-CELL-PC3, and Random Forest signatures, respectively). The signatures provided further stratification for patients with CMS1-3 molecular subtype.Conclusions: The integration of a systems-biology–based biomarker for apoptosis competency with machine learning approaches is an appealing and innovative strategy toward refined patient stratification. The prognostic value of apoptosis competency is independent of other available clinicopathologic and molecular factors, with tangible potential of being introduced in the clinical management of patients with stage III colorectal cancer. Clin Cancer Res; 23(5); 1200–12. ©2016 AACR.

Kaplan-Meier estimates for disease-free and overall survival for n=120 stage III patients of the discovery cohort categorized based on TN staging, tumor location and lymphovascular invasion. Additional exploratory analyses investigated the prognostic value of the APOPTO-CELL-PC3 signature within each sub-group identified by the clinical features.

Unadjusted and multivariate Cox proportional hazards analyses to examine the association of APOPTO-CELL and APOPTO-CELL-PC3 signatures with DFS in n=157 stage III patients of the expansion cohort.

Detailed description of the measurements used as inputs for the apoptosis-based signatures in the discovery, expansion and validation cohorts.

Antibodies targeting the human epidermal growth factor receptor (EGFR) are used for the treatment of RAS wild-type metastatic colorectal cancer. A significant proportion of patients remains unresponsive to this therapy. Here, we performed a reverse phase protein array-based (phospho)protein analysis of 63 KRAS, NRAS, BRAF and PIK3CA wild-type metastatic CRC tumours. Responses of tumours to anti-EGFR therapy with cetuximab were recorded in patient-derived xenograft (PDX) models. Unsupervised hierarchical clustering of pre-treatment tumour tissue identified three clusters, of which cluster C3 was exclusively composed of responders. Clusters C1 and C2 showed mixed responses. None of the three protein clusters showed a significant correlation with transcriptome-based subtypes. Analysis of protein signatures across all PDXs identified 14 markers that discriminated cetuximab-sensitive and -resistant tumours: PDK1 (S241), Caspase-8, Shc (Y317), Stat3 (Y705), p27, GSK-3β (S9), HER3, PKC-α (S657), EGFR (Y1068), Akt (S473), S6 Ribosomal Protein (S240/244), HER3 (Y1289), NF-κB-p65 (S536) and Gab-1 (Y627). Least absolute shrinkage and selection operator and binominal logistic regression analysis delivered refined protein signatures for predicting response to cetuximab. (Phospo-)protein analysis of matched pre- and post-treated models furthermore showed significant reduction of Gab-1 (Y627) and GSK-3β (S9) exclusively in responding models, suggesting novel targets for treatment. This article is protected by copyright. All rights reserved.

Purpose: Apoptosis is essential for chemotherapy responses. In this discovery and validation study, we evaluated the suitability of a mathematical model of apoptosis execution (APOPTO-CELL) as a stand-alone signature and as a constituent of further refined prognostic stratification tools. Experimental Design: Apoptosis competency of primary tumor samples from patients with stage III colorectal cancer (n = 120) was calculated by APOPTO-CELL from measured protein concentrations of Procaspase-3, Procaspase-9, SMAC, and XIAP. An enriched APOPTO-CELL signature (APOPTO-CELL-PC3) was synthesized to capture apoptosome-independent effects of Caspase-3. Furthermore, a machine learning Random Forest approach was applied to APOPTO-CELL-PC3 and available molecular and clinicopathologic data to identify a further enhanced signature. Association of the signature with prognosis was evaluated in an independent colon adenocarcinoma cohort (TCGA COAD, n = 136). Results: We identified 3 prognostic biomarkers (P = 0.04, P = 0.006, and P = 0.0004 for APOPTO-CELL, APOPTO-CELL-PC3, and Random Forest signatures, respectively) with increasing stratification accuracy for patients with stage III colorectal cancer. The APOPTO-CELL-PC3 signature ranked highest among all features. The prognostic value of the signatures was independently validated in stage III TCGA COAD patients (P = 0.01, P = 0.04, and P = 0.02 for APOPTO-CELL, APOPTO-CELL-PC3, and Random Forest signatures, respectively). The signatures provided further stratification for patients with CMS1-3 molecular subtype. Conclusions: The integration of a systems-biology–based biomarker for apoptosis competency with machine learning approaches is an appealing and innovative strategy toward refined patient stratification. The prognostic value of apoptosis competency is independent of other available clinicopathologic and molecular factors, with tangible potential of being introduced in the clinical management of patients with stage III colorectal cancer. Clin Cancer Res; 23(5); 1200–12. ©2016 AACR.

Metabolic reprogramming, rewired signalling cascades, altered microbiota composition and aberrations in the tumor micro-environment have been implicated in CRC progression and response to treatment. In this study, we integrated proteomics, metabolomics, gut microbiota and predictions from network-analysis in tumor and non-cancerous tissue to build a comprehensive atlas to decipher the associations among alterations that occur in the different molecular layers. We performed our molecular fingerprinting in fresh frozen tissue samples from both tumor (centre and/or junction) and matched mucosa (5 and/or 10 cm distance) prospectively collected from a cohort of n=53 stage 0-IV CRC patients. We quantified the protein expression of a panel of 86 cancer-related targets involved in DNA damage/repair mechanisms, cell cycle regulation, growth/angiogenesis signalling, apoptosis and bioenergetics by Reverse Phase Protein Array (RPPA) in n=157 samples (n=48 from tumor and n=109 from mucosa) from n=53 patients. RPPA profiling on both tumor and matched normal tissue was available for 34 out of 53 patients. We performed High Resolution Magic-Angle Spinning Nuclear Magnetic Resonance Spectroscopy (1H HR-MAS NMR) to profile metabolites in n=332 samples from n=52 patients (n=161 and n=171 from tumor and normal tissue, respectively). We characterized gut microbiota features from 16S rRNA sequencing in n=54 samples from n=18 patients. We assessed bioenergetic fitness using an ordinary-differential equations based model of core carbon metabolism that we are developing. The model can simulate the network dynamic and response to fuels (such as glucose, pyruvate and lactate) from extracellular milieu and their utilization and conversion via reactions in the cytoplasm and mitochondrial compartments. We used RPPA-based protein expression for HKII, TIGAR, PGYM, LDHA, MCT4, ATP5A as a proxy for enzymatic activities as case-specific inputs to the model. We identified distinct phenotypic differences when comparing tumors with normal tissue at each data type. Tumors showed higher protein levels for CHK1, GSK3B and LDHA; were enriched with fusobacteria; had increased metabolic levels of glycerphosphocholine, isoglutamine, phosphocholine, taurine and lactate; and were predicted by the mathematical model to have a more glycolytic phenotype. Normal mucosa had higher protein levels for GAB1 and elevated isobutyrate and lipids content. We further analysed the NMR spectra and using a genetic algorithm we identified 10 features (ppm levels) that could classify the tissue type with 88% cross-validated accuracy. We presented a unified analysis connecting dysregulations affecting the proteome, metabolome and microbiome. We provided insights into the heterogeneity of CRC which may lead to more specific molecular classifications and ultimately more targeted treatments. Citation Format: Manuela Salvucci, Liam Poynter, Reza Minerzami, Steven Carberry, Robert O'Byrne, Mattia Cremona, Bryan T. Hennessy, Kirill Veselkov, James Kinross, Jochen H. Prehn. Integrated multi-layer analysis reveals novel insights into the molecular landscape of colorectal cancer (CRC) [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2018; 2018 Apr 14-18; Chicago, IL. Philadelphia (PA): AACR; Cancer Res 2018;78(13 Suppl):Abstract nr 4627.

With 1.4 million new cases every year, colorectal cancer (CRC) is the fourth most common cancer worldwide [Globocan 2012, WHO]. Despite therapeutic advances and improvements in overall care, TNM staging remains the best prognostic indicator for CRC patients’ clinical outcomes and is pivotal for deciding on use of adjuvant chemotherapy after resection of the tumour. Adjuvant chemotherapy is not recommended for many stage II patients and mostly high-risk patients receive chemotherapy. However, there is a lack of robust biomarkers for identifying patient response to chemotherapy, recurrence and mortality risk. We developed a system model of the BCL2 family of proteins (DR_MOMP) to assess the sensitivity of cells to genotoxic stress and to induce apoptosis triggered by chemotherapy. It calculates the stress dose required to induce mitochondrial outer membrane permeabilization (MOMP) based on absolute protein levels and the interaction of pro- and anti-apoptotic BCL2 family proteins. Cells predicted to require a high stress dose showed decreased cell death rates after being exposed to 5FU and Oxaliplatin. Profiles of BAK, BAX, BCL2, BCL(X)L and MCL1 were determined by Reverse Phase Protein Array (RPPA) technology in FFPE primary tumours collected from two distinct cohorts: stage III CRC patients who underwent adjuvant 5FU-based chemotherapy (n = 128), and stage II CRC patients from a completed clinical trial with patients randomised to adjuvant 5FU-based chemotherapy or observation only (n = 138). Protein profiles were inputted into DR_MOMP to determine chemotherapy sensitivity and to classify patients into high- or low risk categories. Findings were validated on the TCGA COAD cohort using both protein (RPPA) and mRNA (SeqV2 RSEM) expression data. Stage II patients classified as high-risk by DR_MOMP and randomised to observation only had approximately 2-fold increased risk of death from CRC compared to those classified as low-risk or received chemotherapy (HR 2.4; 95% CI 1.2-4.8; p-value = 0.0199). Among stage III patients treated with FOLFOX, those classified as high- versus low-risk had a more than 10-fold increased risk of death from CRC (HR 10.6; 95% CI 2.4-46.3; p-value < 0.0001). We validated this finding in 261 stage II-IV patients of the TCGA COAD cohort (HR 10.6; 95% CI 1.2-12.5; p-value = 0.0125). DR_MOMP predicted mortality risk independent of TNM staging and KRAS mutation status. Our system delivers a novel predictive and prognostic biomarker that could be combined with TNM staging when assessing initial risk and subsequent clinical management of CRC patients. Citation Format: Andreas U. Lindner, Manuela Salvucci, Mattia Cremona, Naser Monsefi, Sarah Curry, Clare Morgan, Alexa Resler, Robert O’Byrne, Orna Bacon, Michael Stuehler, Lorna Flanagan, Richard Wilson, Patrick G. Johnston, Manuel Salto-Tellez, Sophie Camilleri-Broët, Deborah A. McNamara, Bryan T. Hennessy, Elaine W. Kay, Pierre Laurent-Puig, Sandra Van Schaeybroeck, Jochen H.M. Prehn. Retrospective evaluation of a system model of the BCL2 family of proteins as a predictive and prognostic biomarker for the clinical outcome of stage II-IV colorectal cancer patients. [abstract]. In: Proceedings of the 107th Annual Meeting of the American Association for Cancer Research; 2016 Apr 16-20; New Orleans, LA. Philadelphia (PA): AACR; Cancer Res 2016;76(14 Suppl):Abstract nr 4924.

Introduction. Somatic mutations in proto-oncogenes and tumour-suppressor genes contribute to rewire the already deregulated metabolic network in cancer cells, resulting in uncontrolled proliferation and oncogenesis. In this study, we set out to establish a dynamic mathematical model of bioenergetics and to exploit it to explore the multifaceted cross-talk between bioenergetics, somatic gene mutations in KRAS and p53, and cell proliferation and survival. Model Development. We have developed an ordinary differential equations-based model of central carbon metabolism in cancer cells which includes glycolysis, pentose phosphate pathway, citric acid cycle and respiratory chain, based on our previous work and published models. The model describes how nutrients (glucose, glutamine, lactate, pyruvate, serine and glycine) from the extracellular micro-environment affect bioenergetics parameters. The resulting model predictions are linked to cell proliferation via a heuristic function. Enzymatic activities regulated by p53 and KRAS mutations were obtained by mining publically available datasets and their regulation by the mutational status was modelled by adapting the corresponding kinetic parameters. To estimate the kinetic parameters, model simulation outputs were fitted to a portfolio of experimental data both generated de novo in house and gathered from the literature in HCT-116 colon cancer cells. Experiments were performed on parental HCT-116 (p53 competent; harbouring a KRAS mutation on exon 2 of codon G13) and three derived mutant cell lines covering all four combinations of p53 and KRAS mutational status to isolate their relative and joint effect on bioenergetics signatures. HCT-116-derived cell lines included: p53 proficient cells with the KRAS allelic mutation silenced by homologous recombination in the presence or absence of p53 knockout by lentiviral shRNA. Results. The model was calibrated against ATP concentrations measured via single-cell microscopy (ATeam probe and TMRM dye) following pharmacological inhibition of respiratory chain complexes (rotenone, sodium azide and oligomycin) as a function of nutrients availability (glucose, lactate, pyruvate). Modelling results revealed that p53 and KRAS mutations drive a shift in metabolic signatures and L-lactate emerged as a pivotal metabolite to stratify the different phenotypes. Systems analysis revealed that in KRAS mutated cells p53 deficiency leads to an increase in glucose uptake and flux through the pentose phosphate pathway and a decrease in lactate production. Indeed, p53 deficient HCT-116 cells showed a decrease in extracellular lactate with respect to p53 proficient cells in validation experiments. Conclusions. The computational model developed can be used to benchmark mechanistic hypotheses by which tumour suppressors and/or oncogenic mutations rewire metabolism and to identify putative targets for therapeutic intervention. Citation Format: Manuela Salvucci, Robert O’Byrne, Natalia Niewidok, Séan Kilbride, Caoimhín G. Concannon, Heiko Düssmann, Heinrich H. Huber, Jochen HM Prehn. Systems analysis of colon cancer cell metabolism rewired by p53 and KRAS mutations. [abstract]. In: Proceedings of the 107th Annual Meeting of the American Association for Cancer Research; 2016 Apr 16-20; New Orleans, LA. Philadelphia (PA): AACR; Cancer Res 2016;76(14 Suppl):Abstract nr 1012.

Colorectal cancer (CRC) is the third and second most commonly diagnosed cancer in males and females, and the second most common cause of cancer-related deaths in the developed world. Identifying the importance of epidermal growth factor (EGF) signalling pathways for the survival of CRC cells (Van Cutsem, Kohne et al. 2011) resulted in development of therapies that target EGF-receptors (EGFR). Late stage CRC patients are mainly treated with monoclonal anti-EGFR antibodies such as cetuximab. Although anti-EGFR therapies have significantly improved survival in CRC patients (Van Cutsem, Kohne et al. 2009), they are not effective in patients with activating KRAS, BRAF, NRAS and PI3KCA mutations (De Roock, Claes et al. 2010). Nevertheless, between 50-60% of patients will not benefit from the additional anti-EGFR treatment, even when these have a quadruple wild-type status (De Roock, De Vriendt et al. 2011), suggesting that novel biomarkers and targeted therapies are required. In this study, we investigate potential biomarkers for cetuximab response in a panel of 93 quadruple wild-type, liver metastatic CRC samples from patient derived xenografts (PDX) with known responses to cetuximab. Matching PDXs were used to measure changes in tumour volumes post cetuximab treatment (Bertotti, Papp et al. 2015) and were later categorised as regressing, progressing or stabilised. Protein expression levels were measured in matched PDXs, using a reverse phase protein array (RPPA) with a panel of 72 antibodies targeting key cancer related proteins as previously described). Statistical analysis of measured values showed significant correlation between high protein expression levels, such as Chk-1, PARP, HIAP-2 (cIAP-1), and PDX progression post cetuximab treatment. These proteins were significantly expressed lower in both regressing and stable PDXs. Interestingly, we found a high cross correlation between expression levels of these proteins across all the samples, giving them a great potential to act as predictive biomarkers for cetuximab response. Bioinformatics pathway enrichment of these proteins showed a significant enrichment of intrinsic apoptotic and cell cycle signalling pathways. These results have also been investigated in publically available patient data with cetuximab response and together will be used to construct a deterministic mathematical cell cycle model that will help to predict the outcome of cetuximab therapy in future. Citation Format: Naser Monsefi, Robert O’Byrne, Steven Carberry, Eugenia R. Zanella, Ana Barat, Mattia Cremona, Clare Morgan, Bryan T. Hennessy, Andrea Bertotti, Livio Trusolino, Jochen H.M. Prehn. Investigating potential molecular biomarkers for cetuximab response in metastatic colorectal cancer tissue using reverse phase protein array. [abstract]. In: Proceedings of the 107th Annual Meeting of the American Association for Cancer Research; 2016 Apr 16-20; New Orleans, LA. Philadelphia (PA): AACR; Cancer Res 2016;76(14 Suppl):Abstract nr LB-354.