DISCUSSION
We present the most comprehensive intra-patient transcriptomic
comparison of tape-strips and biopsies in moderate-to-severe AD to date.
We previously developed a framework for using tape-strips to measure
mRNA and protein in AD in both pediatric and adult populations,
including for monitoring skin activity with topical or systemic
treatment.29,30,69,70 Here, we provided insights into
the common AD phenotype detected by both tape-strips and biopsies as
well as potential differences between these approaches in detecting
specific disease features. Past reports comparing tape-strips and
biopsies in AD are limited either by a small panel of
markers,16,35,36 separation of epidermal and dermal
layers in biopsies,18,37,71 or focus on non-lesional
skin.37,71 An intra-patient, RNA-seq profiling of
active AD lesions comparing tape-strips to full-thickness biopsies, thus
capturing the full spectrum of immune, barrier, and pruritus-related
molecular abnormalities is not available.
Our study showed that both tape-strips and biopsies captured key
molecular immune abnormalities of AD, including T-cell and DC
activation, innate immunity, Th2/Th1/Th17/Th22, and attenuated negative
regulation to varying degrees. However, tape-strips and biopsies also
showed some distinct biomarker profiles, with some notable biomarkers
differentially expressed in only one of the two techniques. Overall
tape-strips outperformed biopsies for most T-helper axes, with greater
differentiation from controls for Th2 (IL-4R, IL-13, IL-31, CCL17,
TNFRSF4/OX40), Th17 (CCL20, CXCL1, CXCL2), innate immunity (IL-1B, IL-8,
TNF), some Th1-related chemokines (CXCL10, CCL3), and negative regulator
markers (IL-34, IL-37). IL-34, in particular, was shown to distinguish
AD from healthy skin with almost perfect accuracy in
tape-strips,30 a property that is likely unique to
tape-strips given the >35x higher fold-change between AD
and controls in tape-strips compared to biopsies. These findings suggest
that chemokines that promote Th2/Th17 differentiation and their defining
cytokines are found in high concentration in the upper epidermis,
whereas their expression may be diluted in larger biopsy samples.
Tape-strips also better captured DC-related products, including myeloid
DCs (ITGAM/CD11B, ITGAX/CD11C), LCs, and inflammatory DCs (CD1A, FCER1A,
CD207/langerin), mainly localized to the
epidermis,48,72,73 which are major sources of
chemokines. Inflammatory DCs correlate with clinical AD severity and
drive Th2 skewing,74,75 representing a potential
pathogenic cell type in AD with better detection by tape-strips.
In contrast, biopsies demonstrated higher fold-changes than tape-strips
for hyperplasia related cytokines and chemokines, including the Th22
cytokine IL-22 and the S100As that are Th17 and Th22
co-regulated,76 suggesting that biopsies may be more
ideal for evaluating hyperplasia markers. IL-22 is primarily expressed
in the dermis during inflammatory states,77,78 while
S100As are produced by suprabasal keratinocytes that are likely below
the depth of tape-stripping.9 Both promote
keratinocyte proliferation and epidermal
hyperplasia.77 IFN-γ, the main Th1 cytokine, and
CCL26, perhaps the best biomarker of therapeutic response to dupilumab,
are additional AD biomarkers primarily localized to the
dermis78,79 and accordingly showed dysregulation only
in biopsies. Markers of an inflammatory dermal fibroblast population
that may orchestrate lymphocyte migration to secondary lymphoid organs
and Th2 immunity, including COL6A5, COL6A6, TNC, and
CCL19,9 were also predictably solely up-regulated in
biopsies. Of note, many markers that are more prominently detected by
tape-strips (IL-4R, IL-13, IL-31, TNFRSF4/OX40) or biopsies (IL-22) are
therapeutic targets for AD that are either FDA-approved or under
investigation,80-88 reinforcing that although both
techniques can detect molecular changes in AD in response to
therapy,80,82,85,86 the type of therapeutic target may
dictate the appropriate sampling technique.
Pruritus, the cardinal clinical feature of AD significantly impacting
patients’ quality of life,89-91 is orchestrated by
complex cross-talk between immune cells, neurons, and
keratinocytes.51,56,63,65,88,92-105 One of the
advantages of tape-strips over biopsies was the ability to better detect
changes in the neuroimmune pathways mediating pruritus. For example, the
Th2 cytokine IL-31, perhaps the key driver of pruritus in
AD,56,88,92-95 was uniquely up-regulated in
tape-strips, with similar trends in OSM and OSMR, which are part of the
same itch pathway and directly interact with IL-31.56STAT3, which mediates IL-31-induced neuronal overgrowth contributing to
skin sensitivity to minimal stimuli,93 was also
dysregulated only in tape-strips. Several transient receptor channel
components (TRPV2, TRPV3, TRPM3) showed more pronounced dysregulation in
tape-strips, likely owing to their expression by sensory neurons, as
well as keratinocytes and/or immune cells.106,107Among these, TRPV3 is most abundant on
keratinocytes,106 and has been shown to be increased
in lesional AD,62 and promote a pro-inflammatory
cascade via the NF-κB pathway.96 In contrast, TRPV6,
which is the only down-regulated TRP gene, with more negative
fold-changes in tape-strips than biopsies, is a pre-requisite for
keratinocyte entry into differentiation.100 As an
example of the complex intercellular interactions that drive itch,
IL-31-induced up-regulation of TRPV3 was proposed to induce keratinocyte
release of SERPINE1, which then directly activates PLAUR/u-PAR in
sensory neurons to perpetuate a cycle of itch and neuroinflammation via
the innate immune system.63,108 Accordingly, SERPINE1
and PLAUR/u-PAR were also exclusively dysregulated in tape-strips.
Overall, given the large contribution of pruritus to the burden of
AD,89-91 the ability to measure the itch signature of
AD before and after treatment through minimally invasive approaches,
such as tape-strips, which in fact outperformed biopsies, is critical.
In addition to immune/neuroimmune dysregulation, tape-strips and
biopsies also preferentially captured distinct aspects of epidermal
barrier impairment. Although markers of terminal
differentiation109-111 (FLG2, SCEL, PSORS1C2) were
decreased in both techniques, at least two-fold greater changes between
healthy and lesional AD were seen in tape-strips. In contrast, biopsies
showed more pronounced up-regulation of markers of epidermal hyperplasia
(KRT6A, KRT16, MKI67), which are primarily expressed in the basal and
immediately suprabasal layers.112, Tight junction
components, which are variably distributed throughout the
epidermis,78 were largely attenuated in AD in both
techniques, with some differences. For example, down-regulation of
CLDN1, which is expressed at all suprabasal layers and has been
described as an AD susceptibility gene with inverse correlations to
Th2,113 was only detected in biopsies. In contrast,
the gap junction components (GJB3, GJB5), which are primarily expressed
in the granular layer,114 were only significantly
decreased in tape-strips. Lipid biosynthesis/metabolism abnormalities,
which are partly driven by Th2 in AD,115 were observed
in both tape-strips and biopsies in line with prior
reports,7,86,115-119 but also showed differences
between the two techniques. For example, genes involved in the
biosynthesis of ceramides, the main lipid constituent of the stratum
corneum (DGS2, SPTLC3),120 were only decreased in
tape-strips, while products related to phospholipid synthesis (AGPAT3,
LPIN1), which are more common in the lower
epidermis,120 were only down-regulated in biopsies.
The inclusion of lipids from the sebocytes of sebaceous glands in
biopsies but not tape strips, which also contribute to barrier
function,121 may further contribute to differences in
lipid-related genes between the two techniques. FAR2, which regulates
synthesis of wax esters exclusively produced in sebaceous
glands,122 and FADS2, which regulates production of
sapienate, the most abundant fatty acid in sebum,123were down-regulated only in biopsies. Overall, tape-strips and biopsies
each preferentially capture gene alterations that complement each other
to constitute the defective barrier phenotype characteristic of AD.
Our study has some limitations. Non-lesional biopsies were not available
for this cohort, preventing a parallel intra-patient comparison of
non-lesional tissues analyzed by both approaches. Nevertheless, prior
studies demonstrated that tape-strips capture the non-lesional AD
phenotype.29,37,69,71,124 Additionally, our analyses
were performed on a primarily adult Caucasian population. Future
analysis should also include other age groups and ethnic backgrounds.
In summary, this is the first global, intra-patient, full-thickness
molecular profiling of lesional skin from moderate-to-severe AD that
compares sampling by tape-strip versus biopsies. Although both capture
the main immune and barrier abnormalities of AD, tape-strips may be the
preferred technique for Th2/Th17, innate immunity, DC, pruritus, and
terminal differentiation related markers. Biopsies may be more
appropriate for evaluating genes below the granular layer, including
markers of epidermal hyperplasia, dermal cytokines (IFN-γ, IL-22), and
inflammatory dermal populations like fibroblasts. The choice of
appropriate sampling technique should be determined considering which
specific immune or barrier features are of greatest importance to the
investigation. Still, this study highlights that tape-strips are not
only adequate but can even outperform biopsies in detecting certain AD
features, including several biomarkers that are therapeutic targets.
Thus, tape-stripping provides a minimally invasive alternative or
supplement to biopsies in clinical trials and longitudinal studies with
the potential to minimize biopsy-related complications and bolster
patient participation in adult and pediatric AD studies.