LC-MS-Based Global Metabolic Profiles of Alternative Blood Specimens Collected by Microsampling

Abstract

Blood microsampling (BμS) has recently emerged as an interesting approach in the analysis of endogenous metabolites but also in metabolomics applications. Their non-invasive way of use and the simplified logistics that they offer renders these technologies highly attractive in large-scale studies, especially the novel quantitative microsampling approaches such as VAMs or qDBS. Objectives: Herein, we investigate the potential of BµS devices compared to the conventional plasma samples used in global untargeted mass spectrometry-based metabolomics of blood. Methods: Two novel quantitative devices, namely, Mitra, Capitainer, and the widely used Whatman cards, were selected for comparison with plasma. Venous blood was collected from 10 healthy, overnight-fasted individuals and loaded on the devices; plasma was also collected from the same venous blood. An extraction solvent optimization study was first performed on the three devices before the main study, which compared the global metabolic profiles of the four extracts (three BµS devices and plasma). Analysis was conducted using reverse phase LC-TOF MS in positive mode. Results: BµS devices, especially Mitra and Capitainer, provided equal or even superior information on the metabolic profiling of human blood based on the number and intensity of features and the precision and stability of some annotated metabolites compared to plasma. Despite their rich metabolic profiles, BµS did not capture metabolites associated with biological differentiation of sexes. Conclusions: Overall, our results suggest that a more in-depth investigation of the acquired information is needed for each specific application, as a metabolite-dependent trend was obvious.

Keywords: blood metabolic phenotype; blood metabolites; blood microsampling (BµS); dried blood spot (DBS); global metabolic profile; liquid chromatography–mass spectrometry (LC-MS); metabolome; quantitative dried blood spots (qDBS); untargeted metabolomics; volumetric absorptive microsampling (VAM).

Figure 1

Sampling of BµS devices used in the study. (A) Dipping a 20 μL capacity Mitra tip into blood sample, (B) pipetting a big drop of blood on a Capitainer device (2 × 10 μL) using a Pasteur pipette, and (C) pipetting 20 µL onto a spot on a Whatman card.

Figure 2

Results of solvent extraction efficiency evaluation for Capitainer (A), Mitra (B), and Whatman (C). (I) Bar chart showing the number of features for each extraction solvent (top) and % of missing features (bottom). (II) Violin plot showing log2 feature intensity distribution for each extraction solvent. (III) Line plot showing number of features at different retention time ranges for each extraction solvent. (IV) Line plot showing log2 of the sum of intensity of features at different retention time ranges for each extraction solvent. (V) Upset plot displaying the unique and common features of each solvent. The first four columns show the number of unique features of the solvent group. The following six columns show the number of features present in two of the compared extracts. The remaining columns represent the features that are common in more than 3 of the extracts analyzed.

Figure 2

Results of solvent extraction efficiency evaluation for Capitainer (A), Mitra (B), and Whatman (C). (I) Bar chart showing the number of features for each extraction solvent (top) and % of missing features (bottom). (II) Violin plot showing log2 feature intensity distribution for each extraction solvent. (III) Line plot showing number of features at different retention time ranges for each extraction solvent. (IV) Line plot showing log2 of the sum of intensity of features at different retention time ranges for each extraction solvent. (V) Upset plot displaying the unique and common features of each solvent. The first four columns show the number of unique features of the solvent group. The following six columns show the number of features present in two of the compared extracts. The remaining columns represent the features that are common in more than 3 of the extracts analyzed.

Figure 2

Results of solvent extraction efficiency evaluation for Capitainer (A), Mitra (B), and Whatman (C). (I) Bar chart showing the number of features for each extraction solvent (top) and % of missing features (bottom). (II) Violin plot showing log2 feature intensity distribution for each extraction solvent. (III) Line plot showing number of features at different retention time ranges for each extraction solvent. (IV) Line plot showing log2 of the sum of intensity of features at different retention time ranges for each extraction solvent. (V) Upset plot displaying the unique and common features of each solvent. The first four columns show the number of unique features of the solvent group. The following six columns show the number of features present in two of the compared extracts. The remaining columns represent the features that are common in more than 3 of the extracts analyzed.

Figure 3

Detected features analysis in BµS and plasma extracts. (I) Bar chart showing number of features (top) and % of missing values (bottom). (II) Violin plot showing log2 of feature intensity distribution for each BµS and plasma. (III) Line plot showing number of features against rt. (IV) Line plot showing log2 of the sum of intensity of features against rt. (V) Upset plot showing common and unique features. The first four columns show the number of unique features. The following six columns show the number of features present in two of the compared extracts. The remaining columns represent the features that are common in more than 3 of the extracts analyzed.

Figure 4

The average intensity of each metabolite in each extract. Color key: red to green = lowest to highest peak area.

Figure 5

PCA scores plot showing all samples analyzed together with the QCs clustered (R2X [1] = 0.513, R2X [2] = 0.148, Q2 cum [1] = 0.394, and Q2 cum [2] = 0.245). BµS was clearly separated from plasma on PC1. BµS was clustered along PC2, with Whatman and Mitra being slightly similar. QC samples were tightly clustered (excluding the first injection) showing our assay’s high precision. Color key: green = C (Capitainer B), purple = M (Mitra), brown = P (plasma), blue = W (Whatman), and yellow = QC.

Figure 6

PCA models of metabolome classification based on sex in the four extracts. (A) PCA in Whatman did not show separation. (B) PCA in Capitainer did not show separation. (C) PCA in Mitra did not show separation. (D) PCA in plasma shows clusters by sex on PC1. Color key: red = females and blue = males.