METABOLOMIC APPROACHES APPLIED TO THE STUDY OF CERVICAL CANCER: A SYSTEMATIC REVIEW

Cervical cancer (CC) is the fourth leading cause of cancer among women worldwide1. One of the possible causes of CC is the persistent infection with Human Papilloma Virus (HPV). HPV is one of the most common sexually transmitted infections in the world, being the main virus associated with the development of neoplasms2 3 4 5.

In recent years the development of new analytical techniques in DNA sequencing and biomolecule measurements has expanded the ability to characterize phenotypic changes in pathological processes6. Metabolomics is capable of simultaneously analyzing multiple metabolic pathways in cells, tissues, and body fluids, connecting genetic processes to phenotypic expression, and listing specific metabolites involved in the pathophysiology of diseases6 7.

Given the current scenario in which CC continues to be a serious health problem for women, metabolomics can be a useful tool in expanding the knowledge on this disease. This review aims to systematically analyze metabolomic approaches in the study of cervical cancer, describing the specific metabolites associated with the development of cervical cancer, identifying possible biomarkers and metabolomic profiles in the context of cervical neoplasia which can lead to innovative clinical approaches.

This systematic review was conducted according to the Preferred Reporting Items for Systematic Reviews and Meta-Analysis (PRISMA)8 guidelines and was registered in the International Prospective Register of Systematic Reviews (PROSPERO) database under number CRD42020215301.

A computerized bibliographic search was carried out in the electronic databases MEDLINE / PubMed, Latin American and Caribbean Literature in Health Sciences (LILACS), Scientific Electronic Library Online (SciELO), Virtual Health Library (VHL) and the databases of gray literature: Google Scholar and Open Gray. As descriptors, the terms in force in the Medical Subject Headings (MeSH) and Health Sciences Descriptors (DeCS) corresponding to “Metabolomics” and “Cervical Cancer” were used. The terms were combined with the Boolean operators “AND” and “OR” (Appendix S1). A manual search was also carried out on the reference list of the selected articles. All searches were carried out from 08/28/2020 to 05/11/2021.

The works were identified by title and abstract by two reviewers (N.D.A. and A.J.S.), who followed the inclusion and exclusion criteria. Studies without abstracts whose title suggested meeting the selection criteria were also selected for analysis. All divergences were resolved by consulting a third reviewer (M.A.M.), who finally defined which articles would be fully read. The selected works were read in full by three authors. Then, the studies were included in the systematic review upon agreement of the three reviewers.

Observational studies that included the use of metabolomics, obtained through biological samples, from patients diagnosed with CC were included in the search. Articles describing the metabolic profiles of tissues and samples and the potential use of metabolomics as a diagnostic and prognostic tool were included. There was no limitation on time or publication language. Studies whose interventions involved specific analysis on drug therapy, carried out on animals, case reports, letters, abstracts, conference proceedings, case studies, review or conglomerates were excluded.

The information extracted from studies included: author, date of publication, geographic origin, title, magazine, type of study, financing, among others. Data on the participants of each work, number of participants, sex, age, use of medications and comorbidities were recorded. The patients in the selected studies were divided into three groups: patients with cervical cancer, patients with intraepithelial lesions and the control group constituted by healthy women (not every selected article described all three categories). Finally, data from metabolomic analysis were collected, such as the method used, the analyzed sample (fluid or biological tissue) and the metabolites found in association with cancer.

Article quality was accessed using QUADOMICS9, an adaptation of the Quality Assessment of Diagnostic Accuracy Assessment (QUADAS)10 for studies using omic technologies.

The age of patients with cervical cancer ranged between 40 and 50 years in the studies listed by this systematic review (Table 1). Such data is compatible with studies by the American Cancer Society that reports the average age at diagnosis of 50 years28. In addition, most studies used mass spectrometry (MS) as the basis for the metabolomic analysis (Figure S2B), the same was observed by Ahmed-Salim et al. (2020) in their systematic review on the use of metabolomics in the study of other types of cancer.

The main metabolites described as relevant by the articles studied in this review are amino acids and components related to carbohydrate metabolism (Table 2). Of the 24 compounds found in common, 14 were amino acids, which were largely reduced in samples from patients with CC (Table 3). These results are further supported by a previous systematic review by Ahmed-Salim et al. (2020) which also observed a decrease in amino acid levels in patients with ovarian cancer.

The decrease in amino acid levels in samples from patients with CC may be related to an increased demand for these metabolites by cancer cells, which need them for survival and growth30. These compounds are linked to several metabolic pathways, participating in anabolic and energetic processes by supplying raw materials such as carbon and nitrogen, in addition to playing an important role in cell signaling pathways30. Figure 1 illustrates the pathways in which the main metabolites described in this systematic review are involved, demonstrating the link between amino acids and energetic pathways. Glutamine is reported as decreased in the blood of patients with CC by the studies listed in this review.

Legend: Metabolites in orange were cited as relevant by three or more articles, the metabolites in green were cited by at least two studies. Metabolites with (*) had a consistent disturbance in all articles in which they were mentioned. Metabolic pathway references extracted from Kanehisa (1996).

The decrease of this component in the plasma may be directly related to the activation of oncogenes and the inactivation of tumor suppressors, such as the Rb protein, which induce the use of glutamine and other important metabolites during the unregulated proliferation of neoplastic cells.

The most prominent altered metabolites within carbohydrate metabolism were β-glucose, α-glucose and lactate. β-glucose levels were decreased in most articles that reported it, in contrast, there was no consensus for α-glucose. Lactate was reported to be increased in three of the four studies that described lactate alteration (Table 3). Turkoglu et al. (2016) also found contradictory results regarding the regulation of carbohydrate related metabolites, in a systematic review of ovarian cancer biomarkers. Taken together these reports support the association between alterations in energy metabolism and cancer. However, the literature points to a lack of consistency in regards to the types and directional regulation of these cancer associated metabolites.

Glucose depletion and lactate increase maybe due to the Warburg effect, which suggests that an anaerobic glycolysis process can occur in tumors where glucose is converted into lactic acid even in the presence of oxygen32. As the tumor expands and the need for oxygen increases, the use of glycolysis becomes an advantageous option, to prevent the development of hypoxic conditions33. In addition, it is believed that this process creates a favorable catabolic state for tumor cells during rapid growth. The more undifferentiated the cancer cell, the greater the lactate production resulting from this process32 34. The Warburg effect is further involved in cell proliferation and malignant progression of cancer through an intricate cell signaling pathway that promotes metabolic changes32 35. Thus, neoplasms in different stages may have different levels of energetic metabolites, which could possibly explain the contradictory findings in the listed studies.

Other metabolites such as phosphatidylcholine, phosphatidylethanolamine, lysophosphatidylcholine and choline were found to be significantly increased in the reviewed studies (Table 3). Wang et al. (2010) described a similar increase in choline levels in a systematic review focusing on colorectal cancer. Choline is an essential nutrient for the synthesis of the plasma membrane. Through the action of choline kinase (CHK), choline is phosphorylated into phosphocholine, a precursor to phosphatidylcholine (Figure 1). These compounds derived from choline contribute to cell proliferation and apoptosis processes, thus, changes in the metabolism of these components can affect cell growth and tumor survival37 38.

The diagnostic capacity of metabolomics was tested by nine studies in this review. Sensitivity was greater than 90%, specificity ranged from 73% to 99% and AUC varied from 0.78 to 0.99 for sets of metabolites. These results demonstrate that metabolomic approaches can be an accurate method for the diagnosis of cervical cancer. Ilhan et al. (2019), Khan et al. (2019), Hasim et al. (2012), Chen et al. (2021) and Cheng et al. (2020) also highlighted the ability of metabolomics to differentiate cancer patients from those with CIN (Table 3). Other systematic reviews have also pointed to metabolomics as a potential method to diagnose ovary29; colorectal36 and endometrium cancers 39. The use of metabolomics to identify specific types of cancer, by evaluating specific differences between metabolites, would be extremely useful for clinical practice. However, there are still no tools or research capable of establishing distinctions between different malignancies based solely on metabolic profiles 7.

The analysis carried out using QUADOMICS (Table S1) exposes a lack of detailed descriptions about patient selection and use of a standard diagnostic test for CC. Likewise, most studies were not blinded to patient diagnosis or did not make this information clear, potentially introducing bias. The heterogeneity of the results obtained using QUADOMICS was an important limitation, although most articles (70.5%) fulfilled at least 10 of the 15 items analyzed, others were rated poorly due to the lack of methodological and/or descriptive rigor in the study methodology. In addition, the different biological samples used, as well as the lack of cancer staging, were other limiting points within our analysis.

The reviewed CC metabolomic studies suggest blood as the main type of biological sample used and mass spectrometry as the preferred analytical platform. The average age of patients with cervical cancer ranged between 40-50 years and most studies added a comparative group of patients with cervical intraepithelial lesions to perform their analysis. Furthermore, we can conclude that the methodological quality of the identified articles was heterogeneous, with a gap of 8 QUADOMICS criteria being met between the studies with the highest scores compared to the lowest scores.

The main metabolites found to be associated with cervical cancer were alanine, creatine, valine, tyrosine, isoleucine, phosphatidylcholine, acetate, lactate and β-glucose. This suggest that a metabolomic profile associated with changes in energy pathways and their connections with amino acid depletion, as well as with changes in the metabolism of glycerophospholipids may be common to CC. In addition, the results described point to the effectiveness of metabolomics in the diagnosis of cervical cancer and in the differentiation of patients with intraepithelial lesions. The expansion and refinement of metabolomic approaches towards CC hold great potential for clinical applications.