Plasma metabolomic profiling of patients with diabetes-associated cognitive decline

Abstract

Diabetes related cognitive dysfunction (DACD), one of the chronic complications of diabetes, seriously affect the quality of life in patients and increase family burden. Although the initial stage of DACD can lead to metabolic alterations or potential pathological changes, DACD is difficult to diagnose accurately. Moreover, the details of the molecular mechanism of DACD remain somewhat elusive. To understand the pathophysiological changes that underpin the development and progression of DACD, we carried out a global analysis of metabolic alterations in response to DACD. The metabolic alterations associated with DACD were first investigated in humans, using plasma metabonomics based on high-performance liquid chromatography coupled with quadrupole time-of-flight tandem mass spectrometry and multivariate statistical analysis. The related pathway of each metabolite of interest was searched in database online. The network diagrams were established KEGGSOAP software package. Receiver operating characteristic (ROC) analysis was used to evaluate diagnostic accuracy of metabolites. This is the first report of reliable biomarkers of DACD, which were identified using an integrated strategy. The identified biomarkers give new insights into the pathophysiological changes and molecular mechanisms of DACD. The disorders of sphingolipids metabolism, bile acids metabolism, and uric acid metabolism pathway were found in T2DM and DACD. On the other hand, differentially expressed plasma metabolites offer unique metabolic signatures for T2DM and DACD patients. These are potential biomarkers for disease monitoring and personalized medication complementary to the existing clinical modalities.

Fig 1. PCA model and OPLS-DA models with corresponding values of R2X, R2Y, and Q2.

(A) PCA score plot of healthy controls (green diamond), T2DM patients (red square), DACD patients (blue circle) and QC samples (yellow triangle); (B) OPLS-DA score plot of healthy controls (green diamond) vs T2DM patients (red square); (C) OPLS-DA score plot of healthy controls (green diamond) vs DACD patients (blue circle); (D) OPLS-DA score plot of T2DM patients (red square) vs DACD patients (blue circle); (E, F, G) Validation plot obtained from 100 tests, respectively.

Fig 2. Comparison of metabolomic profiles from T2DM patients vs healthy controls.

(a) Heat-map of fold change of 56 differential metabolites; (b) Discrimination of T2DM in the training set using Linolenic Acid (1), deoxycholic acid (2), and the combination of them (3). (c) Discrimination of T2DM in the test set using Linolenic Acid (1), deoxycholic acid (2), and the combination of them (3); (d) Scatter plot of Linolenic Acid; (e) Scatter plot of deoxycholic acid.

Fig 3. Comparison of metabolomic profiles from DACD patients vs healthy controls.

(a) Heat-map of fold change of 66 differential metabolites; (b) Discrimination of DACD in the training set (1) and in the test set (2) using phytosphingosine.

Fig 4. Comparison of metabolomic profiles from T2DM vs DACD patients.

(a) Heat-map of fold change of 33 differential metabolites; (b) Discrimination T2DM from DACD in the training set using phytosphingosine (1), sphinganine-phosphate (2), and the combination of them (3). (c) Discrimination T2DM from DACD in the test set using phytosphingosine (1), sphinganine-phosphate (2), and the combination of them (3).

Fig 5. Metabolic network of the significantly changed metabolites within 5 steps by cytoscape software package.

The normalized contents are shown under the chemical name. All the p values were calculated using Student t test., *p< 0.05, **p< 0.01.