Non‐invasive metastasis prognosis from plasma metabolites in stage II colorectal cancer patients: The DACHS study

Abstract

Metastasis is the main cause of death from colorectal cancer (CRC). About 20% of stage II CRC patients develop metastasis during the course of disease. We performed metabolic profiling of plasma samples from non‐metastasized and metachronously metastasized stage II CRC patients to assess the potential of plasma metabolites to serve as biomarkers for stratification of stage II CRC patients according to metastasis risk. We compared the metabolic profiles of plasma samples prospectively obtained prior to metastasis formation from non‐metastasized vs. metachronously metastasized stage II CRC patients of the German population‐based case–control multicenter DACHS study retrospectively. Plasma samples were analyzed from stage II CRC patients for whom follow‐up data including the information on metachronous metastasis were available. To identify metabolites distinguishing non‐metastasized from metachronously metastasized stage II CRC patients robust supervised classifications using decision trees and support vector machines were performed and verified by 10‐fold cross‐validation, by nested cross‐validation and by traditional validation using training and test sets. We found that metabolic profiles distinguish non‐metastasized from metachronously metastasized stage II CRC patients. Classification models from decision trees and support vector machines with 10‐fold cross‐validation gave average accuracy of 0.75 (sensitivity 0.79, specificity 0.7) and 0.82 (sensitivity 0.85, specificity 0.77), respectively, correctly predicting metachronous metastasis in stage II CRC patients. Taken together, plasma metabolic profiles distinguished non‐metastasized and metachronously metastasized stage II CRC patients. The classification models consisting of few metabolites stratify non‐invasively stage II CRC patients according to their risk for metachronous metastasis.

Integrating Two-Dimensional Gas and Liquid Chromatography-Mass Spectrometry for Untargeted Colorectal Cancer Metabolomics: A Proof-of-Principle Study

“…Even if a metabolite was reported in more than one study, often the direction association was different [ 23 ]. For instance, there were no lipid metabolites commonly detected in the same direction in recent studies [ 26 , 28 , 29 ]. The most consistently observed in the same direction as well as by more than one platform were pyruvic acid (up-regulated) and tryptophan (down-regulated) [ 23 ].…”

“…In the past several years, a number of global metabolomic studies on CRC have been published. These studies utilized diverse biological specimens, including blood (plasma/serum) as the most common type [ 22 , 23 , 24 , 25 , 26 , 27 , 28 , 29 , 30 , 31 , 32 , 33 , 34 , 35 , 36 , 37 , 38 , 39 ], followed by tissue [ 27 , 39 , 40 , 41 , 42 ], feces [ 39 , 40 , 43 ], and urine [ 44 , 45 ], while others aimed specifically to detect volatile metabolites in breath and in other types of biospecimens [ 21 , 46 ]. To our knowledge, only two of these were prospective studies where biospecimens were collected before cancer development [ 26 , 34 ], aiming to identify etiological pathways.…”

“…Each study applied heterogeneous criteria for differential abundance between cases and controls, yielding the number of differential metabolites ranging from 5 [ 30 ] to 72 [ 31 ]. More recently published studies identified nearly or over 600 named metabolites [ 26 , 28 ], while another recent paper reported a vast number of differentially present metabolites, i.e., 693, in plasma of CRC cases compared with polyp patients [ 27 ]. It should be noted that the number of annotated metabolites is affected by many factors, including sample treatments, bioanalytical platforms used for analysis, and the methods of data analysis.…”

Beneficiaries of radical surgery among clinical complete responders to neoadjuvant chemoradiotherapy in rectal cancer

“…Risk stratification using non‐invasive parameters has been widely applied in various tumor types for its efficient information provided in clinical decision making. A pretreatment plasma metabolite‐based risk stratification for metastasis in stage II colorectal cancer prognosticated metastasis and non‐metastasis, which provided important information for individualized treatment 40 . Similarly, a smaller sample study combined clinical characteristics and serum protein signatures showed satisfactory synergetic predictive value to group high‐risk and low‐risk patients with pancreatic intraductal papillary mucinous neoplasms 41 .…”

Prognosis prediction of hepatocellular carcinoma after surgical resection based on serum metabolic profiling from gas chromatography-mass spectrometry

“…However, the long-term survival of HCC has not been significantly enhanced [4]. The high recurrence and metastasis rates seriously affect the surgical efficacy and survival of patients, which is an important factor causing poor prognosis of surgical treatment [5,6]. Therefore, effective prediction of the prognostic risk of surgical treatment is particularly important and in urgent need for guiding clinical decisions and Chengnan Fang and Benzhe Su contributed equally to this work.…”