Spatially resolved tissue lipidomics
is essential for accurate
intraoperative and postoperative cancer diagnosis by revealing molecular
information in the tumor microenvironment. Matrix-free laser desorption
ionization mass spectrometry imaging (LDI-MSI) is an emerging attractive
technology for label-free visualization of metabolites distributions
in biological specimens. However, the development of LDI-MSI technology
that could conveniently and authentically reveal molecular distribution
on tissue samples is still a challenge. Herein, we present a tissue
imprinting technology by retaining tissue lipids on 2D nanoflakes-capped
silicon nanowires (SiNWs) for further mass spectrometry imaging and
cancer diagnosis. The 2D nanoflakes were prepared by liquid exfoliation
of molybdenum disulfide (MoS2) with nitrogen-doped graphene
quantum dots (NGQDs), which serve as both intercalation agent and
dispersant. The obtained NGQD@MoS2 nanoflakes were then
decorated on the tip of vertical SiNWs, forming a hybrid NGQD@MoS2/SiNWs nanostructure, which display excellent lipid extraction
ability, enhanced LDI efficiency and molecule imaging capability.
The peak number and total ion intensity of different lipids species
on animal lung tissues obtained by tissue imprinting LDI-MSI on NGQD@MoS2/SiNWs were ∼4–5 times greater than those on
SiNWs substrate. As a proof-of-concept demonstration, the NGQD@MoS2/SiNWs nanostructure was further applied to visualize phospholipids
on sliced non small cell lung cancer (NSCLC) tissue along with the
adjacent normal tissue. On the basis of selected feature lipids and
machine learning algorithm, a prediction model was constructed to
discriminate NSCLC tissues from the adjacent normal tissues with an
accuracy of 100% for the discovery cohort and 91.7% for the independent
validation cohort.