Colorectal cancer: the potential of metabolic fingerprinting

Abstract

Metabolomics is a relatively new omics science that can provide a strong individual small-molecule fingerprint. Disease onset can be monitored as a deviation from the normal healthy fingerprint at the systemic level or in tissues from the diseased organ(s). By applying mass spectrometry and nuclear magnetic resonance as analytical platforms, metabolomics has been used for colorectal cancer phenotyping at different levels. The metabolic profile as a whole is a complex biomarker of diagnostic and prognostic value. Results are promising for the implementation of the method at the clinical level, but larger scale studies and extensive standardization of the pre-analytical phase are needed for a validated definition of the colorectal cancer signature.

Keywords::

• colorectal cancer
• mass spectrometry
• metabolome
• metabolomics
• NMR

The metabolome is defined as the complete set of low-molecular-mass (<1500–2000 Da) compounds in a biological sample [1]. In humans, below this threshold fall largely diverse chemical species that are substrates and by-products of reactions catalyzed by enzymes encoded by the subject’s genome and by the genome of his/her bacterial microflora as well as by molecules introduced via diet, drugs and pollutants. The Human Metabolome Data Base [1] reports more than 40,000 molecules, categorized as: detected and quantified; detected; expected but never detected. The first two categories together include about 50% of the total molecules. Metabolomics could fill the genotype–phenotype gap [2] for its ability to provide snapshots of the functional state of an organism through the measure of the ‘complete set’ of its small-molecule components. ‘Completeness’ is a relative concept: the dynamic range of metabolites concentration is very ample with molecules regarded as abundant if >1 μm or rare if <1 nM [1], and the experimentally accessible fraction of the metabolome depends on the sample type and sensitivity of the analytical platform. The main methodologies are mass spectrometry (MS) and nuclear magnetic resonance (NMR) spectroscopy [3]. MS differentiates metabolites as a function of their mass/charge ratio with very high sensitivity, but requires ad hoc approaches for different classes of chemical compounds, is generally coupled with chromatographic separations and suffers from reproducibility problems in quantitative experiments. NMR, instead, compensates its much lower sensitivity with a number of advantages: the NMR detectable part of the metabolome can be measured in a single shot through a one-dimensional ¹H spectrum with high reproducibility, accurate evaluation of the concentrations and versatility for analyzing metabolites in solution (biofluids, cells or tissue extracts) and in intact tissues, with minimal sample handling.

The accessible part of the metabolome in routinely profiling studies is restricted to the most abundant molecules, from tens to few hundreds. The challenge of metabolomics studies is the ability to extract, via statistical approaches common to other –omics, valuable information from this relatively small and dynamic dataset of biomolecules. A key reference point is the existence of an individual metabolic phenotype that, for healthy subjects, is stable over the time range of years [4,5]. The characteristic phenotype can be extracted as the invariant part of the metabolomic profile and is well defined even when embedded within the day-to-day intrapersonal variability of urine [4]. Deviations from what can be considered the ‘healthy’ metabolic status are of potential interest in disease diagnosis. Metabolic fingerprint has, therefore, become an active field of research for the identification of the signature of various diseases, including cancer, through a global untargeted approach aimed at the identification of possible changes in the levels on the entire set of detectable metabolites, rather than focusing on individual metabolites or on specific metabolic pathways.

Metabolomics has the potential to detect and characterize tumor because alterations to cellular metabolism are a crucial hallmark of cancer [6]. Colorectal cancer (CRC) has very high incidence and bad prognosis, with estimated number of new cases per year in the USA of the order of 140,000 and estimated deaths of the order of 50,000 that settle it as the third most common cancer (Cancer Facts & Figures 2014) and is, therefore, one of the preferred targets in metabolomics [7].

A good share of available studies (∼30%) deals with the characterization of metabolites levels in CRC tissues in comparison with healthy tissues via gas or liquid chromatography MS and/or high-resolution magic angle spinning (HR-MAS) NMR. Metabolomics of tissues is particularly informative because it directly reports the metabolome of the diseased organ, where biomarker variations with respect to a healthy status are expected to be most evident. In tissue analysis, the confounding factors deriving from interpersonal variability can be quenched by comparing malignant and adjacent normal mucosa from the same patient, although intrinsic heterogeneity of tissue samples may require multiple collections and might be one of the sources of the differences in the list of significant metabolites reported by different articles. Each study by itself detects a molecular profile characteristic for CRC, although the lists of metabolites whose levels are altered are never identical [8–10]. Nevertheless, the altered levels always refer to metabolites amenable to main biochemical processes characteristic of cancer cells such as needs for increased energy supply, macromolecule production and the maintenance of redox balance under increased oxidative stress [6]. The differences can be explained by the mentioned differential ability of the various platforms to measure molecules with different chemical properties or abundance as well as by the adoption of different collection procedures. The definition of validated standard operating procedures in sample collection is not trivial: we have recently demonstrated how intraoperative ischemia can sensibly alter the levels of a number of important metabolites [11]; the use of biopsies instead of surgical resection is a valuable option. Metabolic aberration in non-tumor tissue located closer to CRC tissue has been reported [9,10] and the off-tumor mucosa has been proposed to harbor information of possible prognostic relevance [10]. Additionally, a distinct metabolic phenotype has been reported for rectal cancer and colon cancer, which is the possible origin of the differential recurrence rate and metastatic potential [8]. Metabolic fingerprinting via HR-MAS NMR has been suggested ‘as near-real-time surgical investigations, because untreated tissue samples can be analyzed in less than 10 min’ [12]. Metabolomic fingerprinting of different CRC cell lines has highlighted the existence of distinct metabolic patterns attributable to differences in the underlying tumor biochemistry [13]. Metabolic fingerprinting of CRC biopsies may complement the metabolic profiling of tumor cells, providing a better understanding of the environment of the tumor that determines its ability to grow and metastasize. In the long term, the knowledge of tumor microenvironment chemistry might contribute to the development of innovative responsive diagnostic and theranostic agents targeting specific chemical features of the cancer tissues determined by their peculiar small-molecule composition, rather than specific receptors/enzymes.

A major challenge in CRC is non-invasive early diagnosis, and here metabolomics of biofluids comes into play. A defined metabolomic signature of CRC has been detected in a number of studies based on the metabolic fingerprinting of plasma or serum samples of CRC patients with respect to those of healthy controls, reaching discrimination accuracies up to 93.5% [14]. This figure rises up to 100% when comparing healthy subjects and metastatic patients [15]. Within the group of patients with metastatic CRC, statistically significant differences in the serum NMR metabolic profile exist that could be related to their overall survival [15]. Metabolomic profiles are a complex multimolecular measure of how each individual copes with the disease; as such they have a better prognostic value than single-molecule markers and are less subjective than performance status scores like ECOG-PS. In urine, several metabolites contributing to the CRC signature are metabolites deriving from gut microbial-host co-metabolism, allowing to achieve a total area under the curve ≥0.993 [16]. The role of intestinal microbial in CRC onset also deserves attention and deeper future systematic studies are needed to verify the hypotheses that CRC might be a bacteria-related disease [17]. The metabolomic screening of fecal samples, which reflects the status of the full length of the colorectum, might further contribute to the development of non-invasive multimolecular screening tests [18,19]; the proposed contribution of processes beyond the direct shedding of tumor cells to the fecal metabotype [19] might get new impulse from the ‘carcinogenic marriage’ between common flora and intestine [17]. Among metabolites released in the bloodstream, there are volatile organic compounds that might be revealed as very low concentration components of exhaled breath; gas chromatography/MS analyses have shown that the exhalations of CRC patients exhibited differences in their volatile metabolite profiles with respect to healthy subjects opening new routes for non-invasive CRC screening approaches [20].

The power of metabolomic profiling resides in its complexity. A detailed interpretation of the underlying biochemistry might be difficult because each of the detected metabolites is involved in multiple biochemical processes, but the signature of a specific disease gains from concerted changes in the levels of multiple molecules. The overall metabolic profile itself becomes a complex biomarker for diagnosis and design of intervention strategies. At a level of high-throughput screening, NMR has the potential to become the preferred platform for the ease of use, reproducibility and ability to quickly provide a complete profile containing the signature of most abundant metabolites. The accuracy of the approach will largely benefit if single individuals are followed non-invasively over the years to timely monitor deviations from their own healthy metabolomic phenotype. The ability to attribute deviations to a specific disease is determined by the availability of a strong model defining its distinct metabolic signature. As for most diseases, the available articles on metabolomics of CRC are proof-of-principle studies based on patients cohorts of the order of hundreds subjects that need to be validated in larger-scale studies with highly standardized procedures covering both the analytical and pre-analytical steps [11,21]. Needs may be adequately fulfilled through a close collaboration between biobanks, as the main sources of samples in large-scale multicenter studies, and the metabolomics community that may propose and validate the best procedures for successful application of the analytical platforms. There is an increasing effort in the standardization of the analytical phase, which involves research groups and instrument producers, as well as valuable activities in the implementation of metabolomics centers in the clinics for the diffusion of the metabolomics culture in the medical environment.

Financial & competing interests disclosure

The author has no relevant affiliations or financial involvement with any organization or entity with a financial interest in or financial conflict with the subject matter or materials discussed in the manuscript. This includes employment, consultancies, honoraria, stock ownership or options, expert testimony, grants or patents received or pending, or royalties.

No writing assistance was utilized in the production of this manuscript.