Global and targeted metabolomic profiling of colorectal cancer progression

Beatriz Sanchez-Espiridion, Lindsey White, Lopa Mishra, Gottumukkala S. Raju, Scott Kopetz, Jian Gu, Yuanquing Ye, Xifeng Wu and
Dong Liang
Proceedings: AACR 107th Annual Meeting 2016; April 16-20, 2016; New Orleans, LA
• Background: Colorectal cancer (CRC) is one of the most prevalent and deadly cancers in the world. The development of improved and robust
biomarkers to enable screening, surveillance, and early detection of CRC continues to be a challenge. Patients with colorectal adenoma are at
higher risk of developing colon cancer; however, noninvasive methods to identify these patients are still on demand. The aim of this study was
to identify biomarkers of CRC disease progression by using metabolomic profiling of human serum samples in a multistep approach.
• Methods: We performed global metabolomic profiling on 30 human serum samples from patients with colorectal adenoma, 30 CRC patients
and 30 healthy controls who were matched by age, gender and ethnicity. For validation, we measured the three top differentially expressed
metabolites in an additional set of 50 adenoma, 50 CRC and 50 healthy controls.
• Results: Global biochemical profiles of 404 metabolites were detected, with 301 metabolites remaining after quality control procedures. In
discovery phase, 50 metabolites had differential levels between colorectal adenoma, CRC and controls (P for trend <0.05), with 19 metabolites
showing increased levels in CRC and adenoma in comparison to controls and 31 metabolites with decreased levels. Further exploratory
analyses of these metabolites showed a key role for metabolic pathways involving urea cycle, caffeine and galactose metabolism as associated
with CRC progression. The top 3 differentially expressed metabolites (Xanthine, Hypoxanthine and D-mannose) were selected for validation.
Consistent with the discovery phase, CRC cases and adenoma had lower levels of Xanthine than controls (mean ± SD; 9.95 ±0.92 mg/ml vs
10.63±0.97 mg/ml; P<0.001 and 9.87±0.75 vs 10.63±0.97 mg/ml; P<0.001). The same trend was observed for Hypoxanthine (mean ± SD; 10.72
±0.62 mg/ml vs 12.29±1.60; P<0.001 and 10.71±0.61 vs 12.29±1.60 mg/ml; P <0.001) whereas higher levels of D-mannose where observed in
both CRC cases and adenoma when compared to controls (3.32±0.58 mg/ml vs 2.32 ±0.90mg/ml; P<0.001 and 3.32±0.58 mg/ml vs 2.32
±0.90mg/ml; P<0.001). Using the median value of controls as a cut-off point, 94% of the adenoma and CRC cases showed low levels of
Xanthine (Odds Ratio (OR) = 10.47, 95% confidence interval (CI) = 2.63-41.63 for adenoma; OR = 38.76 and 95% CI = 6.58-228.51 for CRC). For
Hypoxanthine, 90% of the adenoma and CRC cases showed low levels (OR = 6.50 and 95% CI = 1.98–21.24 for adenoma; OR = 11.19 and 95%
CI = 3.28-38.21 for CRC). For D-Mannose, all adenoma cases had high levels (OR is not available due to 0 count) and 92% of CRC cases (OR =
15.99, 95% Ci = 4.07-62.88, P< 0.001) had high levels of D-Mannose, compared to 50% of controls having high levels of D-Mannose.
• Conclusions: Our results suggest the potential utility of the identified metabolites as new valuable biomarkers for early detection of CRC.
Staging of colorectal cancer using serum metabolomics with 1HNMR Spectroscopy
Farideh Vahabi 1, Sedigheh Sadeghi 2, Mohammad Arjmand 1, Fatemeh Mirkhani 1, Eshagh Hosseini 2, Mahsheed Mehrabanfar 3, Reza
Hajhosseini 3, Ayda Iravani 1, Parastou Bayat 1, Zahra Zamani

Methodology
Five ml blood was collected from people (16 patients) who were on a liquid diet for at least 48 hr and were referred to the Gastroenterology at
Amir Alam Hospital, Tehran for colonoscopy. One group comprised 8 patients with stage 0 to I colon cancer and a second group made up of 8
patients with II, III and IV stage colon cancer. All stages were confirmed by biopsy.
Bruker DRX-500 NMR spectrometer
Serum metabolomics as biomarkers to differentiate early from metastatic
disease and predict relapse in elderly colorectal cancer (CRC) patients.
Anna Rachelle Austria Mislang, Alessia Vignoli, Samantha Di Donato, Chiara Biagioni, Stefania Vitale, Christopher David Hart, Elena Mori, Dimitri
Becheri, Francesca Del Monte, Claudio Luchinat, Angelo Di Leo, Giuseppe Mottino, Leonardo Tenori, Laura Biganzoli

• Background: Adjuvant chemotherapy improves survival in stage II and III CRC, although a significant proportion of patients
(pts) are cured by surgery alone. Improved risk stratification is required to reduce the number of pts treated unnecessarily,
particularly in elderly populations at risk for higher toxicity. Serum metabolomic profiles may act as biomarkers of residual
(micrometastatic) disease, a prerequisite for relapse, and have been prognostic in other tumor types. This study aims to (1)
identify in elderly pts a metabolomic “signature” for metastatic disease (mCRC) that differentiates it from early disease
(eCRC) and (2) create a metabolomic index for eCRC pts that correlates with clinical outcomes and may be predictive for
relapse. Methods: Serum samples from 105 elderly pts (aged ≥ 70) with CRC, (48 mCRC and 57 eCRC with ≥ 5 years follow
up) were pooled from 4 previous clinical trials. Samples were analyzed via Proton Nuclear Magnetic Resonance (NMR) and
the spectra were used to characterize the metabolic profiles of the two cohorts. Principal component analysis (PCA) and
canonical analysis (CA) were applied to obtain the supervised separation of eCRC and mCRC spectra. For the purpose of
classification, the K-nearest neighbors (k-NN) method was applied to the PCA-CA scores. Wilcoxon test was then used to
compare the levels of 34 quantified metabolites between eCRC and mCRC pts. A model that assigns a risk score is being
built based on the degree to which an eCRC serum profile resembles the metastatic profiles. This risk score will be
compared to the Adjuvant!Online estimated risk of relapse score and the actual outcome. Results: PCA-CA-kNN
classification of NMR spectra was able to discriminate eCRC and mCRC with an accuracy of 74%. Four metabolites (2-
methylbutyrate, 2-methylsuccinate, histidine and formate) were found to differ significantly (p < 0.05) between eCRC and
mCRC metabolomic profiles. Conclusions: NMR metabolomic profiles can discriminate early and metastatic CRC. A model to
assess the likelihood of relapse, based on the degree to which an eCRC serum profile resembles the metastatic profiles, is
being built, and results will be presented at the meeting.
Global and Targeted Serum Metabolic Profiling
of Colorectal Cancer Progression
Yin Long, MD, MS1,2; Beatriz Sanchez-Espiridion, PhD1; Moubin Lin, MD, PhD2; Lindsey White, PhD3; Lopa Mishra, MD4;
Gottumakkala S. Raju, MD4; Scott Kopetz, MD, PhD5; Cathy Eng, MD5; Michelle A. T. Hildebrandt, PhD1;
David W. Chang, PhD1; Yuanqing Ye, PhD1; Dong Liang, PhD3; and Xifeng Wu, MD, PhD

Methodology
Group1- 30 healthy controls, 30 PLP patients, and 30 CRC patients for metabolomic profile
screening in the discovery phase
Group2- 50 healthy controls, 50 PLP patients, and 50 CRC patients in the validation phase
Histologically confirmed and recruited at The University of Texas MDAnderson Cancer Center
between 2006 and 2013
LCMS