nature reviews genetics https://doi.org/10.

1038/s41576-022-00553-x

Review article Check for updates

Spatial biology of cancer evolution

Zaira Seferbekova , Artem Lomakin
1,4
, Lucy R. Yates
1,2,4 3,5
& Moritz Gerstung 1,5

Abstract Sections

The natural history of cancers can be understood through the lens Introduction

of evolution given that the driving forces of cancer development are Tissue organization controls
mutation and selection of fitter clones. Cancer growth and progression evolution

are spatial processes that involve the breakdown of normal tissue Mapping the cancer
ecosystem
organization, invasion and metastasis. For these reasons, spatial
patterns are an integral part of histological tumour grading and staging Evolution of the cancer
ecosystem
as they measure the progression from normal to malignant disease.
Translational opportunities
Furthermore, tumour cells are part of an ecosystem of tumour cells
and their surrounding tumour microenvironment. A range of new Conclusions and future
perspectives
spatial genomic, transcriptomic and proteomic technologies offers
new avenues for the study of cancer evolution with great molecular
and spatial detail. These methods enable precise characterizations
of the tumour microenvironment, cellular interactions therein and
micro-anatomical structures. In conjunction with spatial genomics,
it emerges that tumours and microenvironments co-evolve, which
helps explain observable patterns of heterogeneity and offers new
routes for therapeutic interventions.

Division of AI in Oncology, German Cancer Research Centre DKFZ, Heidelberg, Germany. 2European Molecular
1

Biology Laboratory, European Bioinformatics Institute (EMBL-EBI), Hinxton, UK. 3Wellcome Sanger Institute,
Hinxton, UK. 4These authors contributed equally: Zaira Seferbekova, Artem Lomakin. 5These authors jointly
supervised this work: Lucy R. Yates, Moritz Gerstung. e-mail: moritz.gerstung@dkfz.de

Nature Reviews Genetics | Volume 24 | May 2023 | 295–313 295

Review article
Introduction
The drivers of evolution are mutation and selection1,2. Mutation of
somatic cells is an inevitable and persistent consequence of life3–5.
In most normal tissues, mutations accumulate at a steady rate of around
15–50 per cell per year of life, with only the germline known to exhibit
lower rates4. Tumour cells often exhibit an elevated mutation burden
due to the variety of mutational processes they have been exposed to
during life and, in some cases, due to acquired hypermutation6. The con-
tinuous accumulation of mutations inevitably leads to diversification
at the level of single cells both in tumours and normal tissues.
The second force of evolution is selection, which describes how a
fitter lineage outgrows its relatives. Selection operates at the level of
a wide range of heritable phenotypes that derive from genomics and
epigenomics. Clones with a selective advantage in the environment to
which they are exposed will expand, while those with a disadvantage
will tend to disappear (Fig. 1a). Advantageous variants are therefore
enriched in genomes of aged normal tissues and cancers7,8. More than
500 so-called cancer driver genes have been reported to date9–11, which
are believed to cause different cancer hallmark traits and enabling char-
acteristics that involve cell-intrinsic mechanisms as well as interactions
with the tumour microenvironment (TME)12,13.
While mutation continuously generates subclonal diversity at the
level of single cells, it is well documented that tumours are mosaics
of subclones each comprising hundreds of thousands of cells that have
arisen from a shared ancestor14–18 (Fig. 1a); these patterns have been
further confirmed by single-cell studies19–24. The mechanisms of these
expansions remain debated. It has been shown that cancer subclones
exhibit signs of positive selection and driver gene mutations that can
also be found clonally18. However, it is also conceivable that subclones
branching at early stages of tumour development reach considerable
size without a selective advantage as they continue to expand with the
same tumour, a phenomenon termed neutral evolution25.
Subclonal mosaicism is often found to be spatially variegated
as demonstrated in diverse studies based on tumour macrodissec-
tion26–33 and laser-capture microdissection21,34–40 (Fig. 1b). These clonal
variegation patterns were shown to be accompanied by differential
gene expression in multiple cancer types, including renal27, colorec-
tal41, lung42 and breast cancer43, using spatially resolved genomic and
transcriptomic analyses, indicating the presence of subclone-specific
gene expression and associations between certain subclones and
characteristic TMEs.
The exact mechanisms creating these phenomena are not fully
understood, largely owing to a dearth of methods that can molecularly
characterize whole tumour sections with spatial resolution. A variety
of non-exclusive explanations have been proposed. A rapid expansion
could lead to a fragmentation of closely related clones44 even though
it has also been argued that cell dispersal is a crucial element shaping
tumour structure45,46. However, it is also possible that tissue micro-
anatomy, such as the ductal system of the breast, provides a template
for clonal segregation (discussed further in the section ‘Tissue organi-
zation controls evolution’). Lastly, it is conceivable that locally distinct
microenvironments or niches favour the selection of particular clones.
Various new spatially resolved genomic, transcriptomic and pro-
teomic technologies offer novel insights and answers to these unre-
solved questions (Fig. 1b). The technological aspects enabling the
nascent field of spatial cancer biology are summarized in Box 1 and have
been extensively reviewed elsewhere47. Individual technologies differ
in the multiplexity of their readouts and their spatial resolution as well
as in their sensitivity and available field of view. Combining genomic,
Nature Reviews Genetics | Volume 24 | May 2023 | 295–313
transcriptomic and proteomic layers enables a rich characterization
of tumour ecosystems.
In this Review, we focus on the spatial aspects of cancer evolu-
tion, summarize how spatial transcriptomics and proteomics reveal a
complex landscape of the tumour ecosystem, discuss how interactions
with the TME are integral to cancer evolution, and propose potential
clinical applications and future directions. We focus particularly on
the role of tissue micro-anatomy in controlling the rate of evolution
as well as cellular interactions with the TME within which cancer cells
evolve.
Tissue organization controls evolution
The rate at which malignant clones emerge and spread through
the tissue is controlled by its organization. Levels of organization
include the micro-anatomical tissue architecture as well as differen-
tiation hierarchies with few stem cells feeding successively larger pools
of differentiated cells. The breakdown of these protective principles
is a key feature of cancer development. However, the resident tissue
structure, for example, the ductal system in the breast or epithelial
layers of the oesophagus, can also influence the rate of progression
at pre-invasive stages. The transformation ends with overwhelming
metastatic disease.
Normal tissue organization suppresses evolution
It has long been hypothesized that somatic tissues are structured to
suppress the rate of somatic evolution1. These ideas are based on fun-
damental theoretical considerations that the rate of evolution depends
on population structure. Generally, the rate of evolution is determined
by the rate at which new mutations are generated (a product of the
number of cells and the mutation rate per cell), the probability that
new mutants sweep through the population48 and the time it takes to
do so49,50. In addition to suppressing the mutation rate, differentiation
hierarchies can reduce the chance of emergence and slow the spread
of mutant lineages. The haematopoietic system constitutes a basic
example51: its 1013 cells derive from around 100,000 haematopoietic
stem cells52,53, which themselves divide slowly and produce a range
of faster-dividing differentiated cells54. This hierarchy reduces the
rate of evolution, as only mutations in the comparably small number
of stem cells and potentially the first progenitors, or mutations that
lead to a differentiation blockage, will remain in the population. The
colon extends these organizational principles by its compartmen-
talization into micro-anatomical crypts; in each crypt, a similar dif-
ferentiation hierarchy operates in which a very small number of stem
cells replenishes the colonic epithelium54. As differentiating cells are
fated to die within ~5 days, only mutations arising in the stem cells per-
sist1,55,56. Furthermore, micro-anatomy largely constrains the fixation of
stem-cell mutations to individual crypts (Fig. 2a).
Systematic studies of normal tissues confirm that, in tissues with
a more complex architecture, clones typically remain small in size
and develop independently in comparison to less structured tissues
where clonal expansions tend to be larger5. In a recent investigation of
517 normal colonic crypts, only 1% harboured cancer driver gene muta-
tions, and those that were mutated were confined to single crypts57.
Similar observations have been made for glandular tissues such as the
endometrium58: although most of the glands harboured driver muta-
tions, they were limited to a single mutation or a few neighbouring
mutations. Conversely, it is well established that the haematopoietic
system harbours mutant lineages whose frequencies can reach up
to 100%59,60. Other epithelial tissues, including the skin, oesophagus
296
Review article

a Diagnosis

Subclonal frequencies
Cancer
founder
Zygote cell

Driver mutations

500 mutations
Evolutionary time

b Spatial biology Single cell

Genomics

Spatial subclonal Subclonal Single-cell genotypes,

variegation, early interfaces latest evolution
subclonal evolution

Anatomy

Cancer site Organ anatomy Micro-anatomy Cytomorphology

Microenvironment

Spatially segregated Interactions Cell typing

microenvironments
1 cm
Fig. 1 | Spatial biology of cancer evolution. a, Somatic evolution and cancer
development. Human cells accumulate mutations throughout their lifetime.
A subset of these, so-called driver mutations, are associated with neoplastic
transformation. Ongoing mutation and selection cause subclonal divergence within
between cells
1–0.1 mm 10 μm
the tumour clone. b, Spatial technologies add new layers of information. A tumour
is a mosaic of subclones with distinct micro-anatomy and tumour microenvironment.
Spatial analyses offer new avenues for charting and understanding the biology and
evolution of the cancer ecosystem at scales ranging from a single cell to whole tissue.

or bladder, have an intermediate degree of tissue structure and are the haematopoietic system because the dynamics are confined to
patchworks of microscopic clones4,61–63 (Fig. 2a). In such tissues, the two dimensions, which reduces the interfaces at which a fitter variant
fixation probability is high but the time to fixation is greater than in may replace less-fit competitors. Of note, not all mutations causing

Nature Reviews Genetics | Volume 24 | May 2023 | 295–313 297

Review article

Box 1

Enabling technologies
A range of spatially resolved approaches for mapping DNA, RNA and
proteins has been developed47,215. Higher spatial resolution typically
implies smaller molecular quantities (figure, part a). Other important
parameters are the field of view, which ranges from millimetres
squared to centimetres squared, and method sensitivity.
Macrodissection and microdissection
The foundations were laid by macrodissection followed by multi-omic
profiling but new technologies offer a microscopic and single-cell
resolution (figure, part b). In multi-region sequencing27,30, the collection
of several biopsy samples from multiple geographically separated
tissue regions is coupled with later bulk sequencing of DNA or
RNA. Laser-capture microdissection (LCM)216 allows the profiling
of even smaller regions of interest using laser ablation. LCM is very
versatile as it can be combined with high throughput sequencing,
including whole-genome sequencing, and single-cell RNA or DNA
sequencing21,217–219.
Spatial molecular barcodes
This set of methods utilizes probes with unique barcode sequences
to capture molecules and then map them to specific spatial locations
(figure, part c). So far, these technologies have mainly focused on

a MIBI, ImmunoSABER MIBI-TOF Stereo-seq b Macrodissection and microdissection

DNA, epigenetics
10–7
CODEX RNA
corrFISH Seq-Scope
epiMERFISH Protein
Base Scope DVP, DNA-MERFISH
RNAScope BaSISS STARmap, MERFISH DNA mutations via RNA
Resolution, m

10–6 FISSEQ
osmFISH ISS seqFISH+ HDST
IMC smFISH Subcellular
HybISS TSCS, resolution
pciSeq slide-DNA seq
10–5 Typical size of
a human cell
GeoMX Visium Visium HD,
Slide-seq, Supercellular
CosMX resolution
10–4 ST DBiT-seq Multi-regional sequencing LCM based
1 10 102 103 104 105 106 107 108 109
Plexity
Whole transcriptome Whole genome

c Molecular barcode-based methods d Image-based methods e Mass spectrometry

Spot based Beads based FISH based Multiplexed IHC

Type of assay Targeted Untargeted (whole) Type of molecule detected RNA DNA Protein

BaSISS, base-specific ISS; CODEX, co-detection by indexing; corrFISH, correlation FISH; DBiT-seq, deterministic barcoding in tissue for
spatial omics sequencing; epiMERFISH, epigenetic MERFISH; FISSEQ, fluorescent ISS; HDST, high-definition spatial transcriptomics; hybISS,
hybridization-based ISS; ImmunoSABER, immunostaining with signal amplification by exchange reaction; MIBI-TOF, MIBI by time of flight;
osmFISH, ouroboros smFISH, pciSeq, probabilistic cell typing by ISS; smFISH, single-molecule FISH; ST, spatial transcriptomics; STARmap,
spatially resolved transcript amplicon readout mapping; Stereo-seq, spatial enhanced resolution omics sequencing; TSCS, topographic
single-cell sequencing; Visium HD, Visium high definition.

Nature Reviews Genetics | Volume 24 | May 2023 | 295–313 298

Review article
(continued from previous page)
transcriptomic profiling with spatial transcriptomics125, 10x Visium
and Slide-seq220,221 being the most popular protocols among many
others222–224. Spatially resolved DNA sequence capture37 has enabled
spatially resolved genome-wide copy number calling along with
several new methods for spatial epigenomic profiling225,226. Although
these methods support unbiased discovery and result in data with
high molecular content, the resolution is limited by the diameter of
each spot and currently does not allow profiling at the single-cell
level or point mutation detection. Further improvements in resolution
are currently being developed222,224,227.
Image-based technologies
This group of methods involves in situ hybridization of fluorescently
labelled probes (FISH) to nucleic acids of interest to identify their
location with single-cell or subcellular resolution (figure, part d).
A variety of multiplexed extensions228–238 exists, including multiplexed
error-robust FISH (MERFISH)239,240, sequential FISH (seqFISH)241,
RNAscope242 and in situ sequencing (ISS)243 for spatial transcriptomic
profiling. The latter two protocols have been further adapted to
permit mutation detection using the Base Scope244 and BaSISS
a selective advantage are associated with malignant transformation.
For example, in the skin and oesophagus, NOTCH1 mutant clones
expand and suppress malignant evolution while maintaining normal
tissue function61,62,64–66.
Breakdown of evolutionarily confining tissue architectures in
cancer
The fundamental observation that dedifferentiation and the loss of
normal tissue architectures are correlated with cancer aggressiveness
was made a century ago and still forms part of modern histopatho-
logical grading systems that direct cancer treatment67–70. In these sys-
tems, higher grades correspond to more aggressive tumours, typically
composed of less differentiated cells.
The breakdown of such evolutionarily confining tissue structures
is a hallmark of cancer. It is especially noticeable among carcinomas,
which derive from epithelial tissues (Fig. 2b). At the earliest stages
of progression, so-called carcinomas in situ are still restrained to the
resident tissue structures, such as the duct of the breast. Similar to
the considerations of normal tissues, the resident tissue structures may
delay clonal sweeps. This tissue-based restraint was observed in a case
of breast cancer, where spatial genomic analysis revealed two distinct
subclones occupying a checkerboard pattern of micro-anatomical
niches within the ductal system43.
During invasion, tissue architectures are lost, probably via different
mechanisms in each tissue type, but a frequent property of carcinomas
is the epithelial-to-mesenchymal transition (EMT). EMT enables epi-
thelial cells to lose their polarity and escape from the confinement of
planar epithelial tissue organization (as reviewed in ref.71). In colorectal
adenocarcinoma, EMT can be induced via the WNT signalling pathway
by loss-of-function mutations in APC or somatic mutations of CTNNB1
inhibiting its degradation72. Both alterations are believed to be a key
step in the development of adenomas, the neoplastic precursor to
invasive adenocarcinoma. A further step of invasion is the breakdown
and remodelling of the extracellular matrix (ECM), thought to also be
Nature Reviews Genetics | Volume 24 | May 2023 | 295–313
protocols43. Immunohistochemistry (IHC) or immunofluorescence
uses antibodies to localize proteins or other molecules. Highly
multiplexed spatial proteomics with dozens of targets can be
achieved using DNA barcodes245,246, multiplex immunofluorescence
(CyCIF)247 and other approaches248–250. Multi-omic approaches, such
as GeoMx or CosMx251,252, allow researchers to assess multiple layers
of biological information within one cell simultaneously. Image-based
technologies provide high-resolution tissue profiling but require
a pre-defined panel of targets.
Mass spectrometry
A further class of methods combines metal-conjugated antibodies
and mass spectrometry (figure, part e). Spatial resolution is provided
by spatial laser ablation (imaging mass cytometry (IMC))253 or ion
beams (multiplexed ion beam imaging (MIBI))254. These approaches
detect proteins and smaller molecules, such as lipids, metabolites,
and drugs, in single cells253, as reviewed in ref.255. Deep Visual
Proteomics (DVP) combines artificial intelligence-driven LCM
with mass spectrometry to map protein abundances onto tissue
slides256.
facilitated by the microenvironment (see the section ‘Mapping the
cancer ecosystem’).
The consequence of this breakdown is likely the acceleration of
malignant evolution as the fixation probability and the rate of fixation
are higher in unstructured populations with a greater degree of mix-
ing. Several theoretical investigations have demonstrated that spatial
constraints can control the mode of tumour evolution73,74. Another
consequence is that different selective pressures apply, which may be
one cause of the characteristic sequences of cancer progression such
as the classic APC-KRAS-TP53 model for colorectal carcinoma75,76 and
the four stages of TP53 inactivation in the progression of pancreatic
ductal adenocarcinoma from a pre-malignant to malignant state77.
In this light, a pan-cancer analysis revealed that the portfolio of late
and subclonal driver gene mutations tends to be more diverse than
the set of early alterations76.
Metastatic dissemination
The ability of cancer cells to leave a primary tumour site, seed in distant
organs and generate metastatic deposits presents landmark stages
that are strongly predictive of poor clinical outcomes78. This inherently
spatial process involves, upon tissue invasion, intravasation of single or
small aggregates of cancer cells, circulation through the blood system,
arrest in the capillaries of the target organ, extravasation, and prolifera-
tion79 (Fig. 2b). These steps thus involve a range of micro-anatomical
bottlenecks and possible differential selective pressures in different
stages and target organs.
Several multi-region genomic sequencing studies have investi-
gated the phylogenetic relationships of primary tumours and metas-
tases from the same patient27,28,30,32,80–85. These data show that most
metastases have diverged from the primary tumour clone typically
shortly before or after its most recent common ancestor, which is
often dated 1–5 years before diagnosis83,84,86,87. The origin of meta­
stasis appeared to be mostly clonal88–90 even though, in advanced
disease, intermixed subclones have been found in multiple metastases,
299
Review article
a Normal tissue organization
Population structure constraints

Shedding Colonic crypt

Differentiated Wild-type
cell mutates differentiated
cell Mutations in
Mutations in
differentiated stem cells are
cells are lost maintained and
as they get may transform the
replaced by crypt if all other
Differentiation wild-type cells stem cells are
also replaced
Stem cell
mutates
Wild-type stem cell

Physical architecture constraints

Colon Skin Blood

Subclones are confined within Clones and subclones are HSC

single crypts mosaically distributed

Crypts Planar epithelium Well mixed subclones

Slow evolution, small subclones Fast evolution, large subclones

More constrained Less constrained

b Carcinogenesis Tumour cells from different subclones

Without metastatic With metastatic
Normal tissue potential potential

Normal cells

Basement membrane
ECM Stroma
Intravasation of subclones with metastatic
potential into a lymphatic vessel Metastases in lymph nodes

Primary tumour
Polyclonal and
metastasis-to- Monoclonal
metastasis seeding seeding

Basement
membrane
breach Invasion into Intravasation of subclones with metastatic
surrounding Metastases in other organs
potential into a blood vessel
normal tissue
Vascularization
Possible recolonization of the primary site

Nature Reviews Genetics | Volume 24 | May 2023 | 295–313 300

Review article
Fig. 2 | Spatial cancer evolution. a, Normal tissue structure suppresses somatic
evolution. Top: in addition to physical constraints, the differentiation hierarchy
in the colonic crypt limits the number of cells with fixation potential, which
reduces the chances of malignant transformation. Bottom: highly structured
tissues, such as the colon, slow evolution by constraining the spread of clones.
Physically unstructured tissues, such as the haematopoietic system, exhibit the
fastest rate of evolution and clones, which replace the entire organ. b, Spatial
steps of cancer progression. Normal tissue structure breaks down during tumour
progression. This process is exemplified here by a carcinoma: breach of the
suggesting spread from one metastasis to another84,86,91. Pan-cancer
analyses of metastatic cancers have revealed that metastases exhibit
fewer intra-metastatic subclones than primary tumours92–94, which may
in part be a consequence of sampling fewer cells (often core needle biop-
sies). From the perspective of driver gene mutations, it has been argued
that treatment-naive metastasis has broadly the same driver gene muta-
tions as the matched primary95. Characteristics of metastatic cancer
include high levels of genomic instability, TP53 mutations and whole-
genome duplications, shared by both the primary tumour and meta­
stases92,94. However, adjuvant therapies can select for specific mutations,
for example, prostate and breast cancers treated with endocrine depri-
vation therapies are enriched for resistance mutations in AR and ESR1,
respectively93,94.
Although it emerges that tissue architecture has a profound
influence on somatic evolution, it is important to realize that cellular
interactions with normal cells are also likely to affect the fate of cancer
clones. The composition of the TME and its role in cancer evolution will
be discussed in the next two sections.
Mapping the cancer ecosystem
The cellular composition of tumour masses is highly variable, with the
neoplastic lineage contributing as few as 10% of cells in some cases96.
The remaining cells derive from the TME, which is usually dominated
by immune and stromal cells in addition to vessels, adipocytes and peri-
cytes. Together, these cells and other physical structures, such as ECM
and chemical gradients, form an environment termed the cancer eco-
system97. Over the past decade, single-cell RNA sequencing (scRNA-seq)
analyses have catalogued the cellular composition of the cancer eco-
system98–105. More recently, spatial transcriptomic and proteomic tech-
nologies have started to uncover how these diverse cell types organize
into communication interfaces and niches106,107 (Boxes 1 and 2).
Cellular composition of the TME
The TME has been studied extensively and, therefore, we herein only
present a high-level summary of major cell types. Inflammation is a
hallmark of cancer108 and reflects the interplay of antitumorigenic and
pro-tumorigenic immune responses (Fig. 3a). The antitumour response
is primarily generated by cytotoxic CD8+ T lymphocytes that recognize
antigens presented on the tumour cell surface109,110. The success of this
response is dependent on priming and prior education by professional
antigen-presenting cells (dendritic cells) and is facilitated by CD4+
T helper cells111,112. B cells have diverse functions, including antigen
presentation, antibody production and cytokine-mediated cytolytic
effects113. In many cancers, B cells form dense aggregates with T cells
and dendritic cells to form tertiary lymphoid structures114 that are
generally associated with favourable cancer outcomes and responses
to immune-checkpoint blockade115. Natural killer cells might have a
particularly important role in anticancer immunity due to their ability
to recognize stressed cells irrespective of neoantigen presentation116.
By contrast, CD4+ regulatory T (Treg) cells can suppress the immune
response through cytokine release (for example, IL-10) or by modula-
tion of antigen presentation functions. In certain situations, some
Nature Reviews Genetics | Volume 24 | May 2023 | 295–313
basement membrane is followed by tumour invasion into the neighbouring
stroma, lymphangiogenesis and angiogenesis. Through blood and lymphatic
vessels, cancer cells with metastatic potential can transfer to lymph nodes
and distant organs, giving rise to metastases at these new sites. Metastases
containing cancer cells of different lineages within the same lesion are
called polyclonal, whereas metastases containing cancer cells of the same
lineage are called monoclonal. Metastatic subclones may later return to the
primary site and recolonize it. ECM, extracellular matrix; HSC, haematopoietic
stem cell.
immune cells, such as macrophages, myeloid-derived suppressor
cells and granulocytes, can actively promote tumour progression
through two-way interactions with the tumour, other immune cells
and the stroma117–119. The stroma also plays an active role in shaping
the functional immune response to cancer. A variety of studies have
demonstrated that cancer-associated fibroblasts and myofibroblasts
typically change shape, become more proliferative, undergo diverse
transcriptional and metabolic changes, engage in angiogenesis, and
interact with leukocytes and tumour cells while the resulting ECM
changes in chemical composition and mechanical properties120–124.
Cellular interactions and neighbourhoods
Although scRNA-seq studies have generated comprehensive catalogues
of detailed TME cell types and states, these data only allow circumstan-
tial insights into how cells interact. Spatial transcriptomic and prot-
eomic technologies are beginning to show that the cells constituting
the cancer ecosystem self-organize into micro-anatomical neighbour-
hoods, which could provide insights into the frequency and nature
of different cellular interactions (Fig. 3b–d and Box 2). The majority of
spatial omic cancer studies to date have used spatial transcriptomics
as commercialized by 10x Visium (Box 1). These studies reveal spatially
structured TMEs in breast125–127, prostate128, pancreatic129, colorectal130,
skin131, brain132, liver133, bladder134 and other cancer types135,136. The ability
to use whole-transcriptome readouts also helped establish the presence
and broad co-localization of rare cell types (Fig. 3b). However, the cur-
rent super-cellular resolution (55 µm, ~3–10 cells) requires informatic
deconvolution of the measured signals in each spot.
Highly multiplexed antibody detection approaches have been
used to generate single-cell resolution readouts of the spatial distribu-
tion of the major cell types in breast cancer, colon cancer and mela-
noma, amongst others137–140. These approaches permit the detection
of microscopic neighbourhoods with distinct composition of tumour,
immune, stromal and vascular cells. In two studies that analysed hun-
dreds of breast cancers by applying imaging mass cytometry (Box 1)
to tissue microarrays, several recurrent neighbourhoods were identi-
fied137,138 (Fig. 3c). The biological feasibility of the identified neighbour-
hoods was provided by their relationship with well-defined clinical
subtypes and survival outcomes. For example, consistent with a failing
antitumorigenic response, ‘suppressed expansion structures’ con-
tained many dysfunctional T and Treg cells and were enriched in breast
cancers with poor clinical outcomes. By contrast, a ‘vascular stroma’
signature was enriched in a very favourable outcome subgroup of hor-
mone receptor-positive breast cancers (Fig. 3c). The spatial patterns
formed by individual cell types were also examined. In breast cancers,
whereas T cells were diffusely scattered, B cells tended to form aggre-
gates, were in multiple neighbourhood types, and were associated with
disparate clinical subtypes and clinical outcomes, suggesting that B cell
function might be dependent on the precise TME context141 (Fig. 3c).
However, some tumour phenotypes, such as cell proliferation, can be
shaped by further long-range interactions142.
The proximity of cells does not necessarily equate to an interac-
tion. To address this limitation, a combination of CyCIF (Box 1) and
301
Review article

Box 2

Spatial omic analysis

A typical tumour cross-section covers several centimetres of tissue
containing up to 1,000,000 cells. Spatial omic methods characterize
the molecular features of these cells with a resolution as fine as
~0.2 μm, depending on the method. The wealth and detail of data
bring a series of data analysis challenges257 (see the figure).
Raw signal analysis
In the case of molecular barcode-based technologies (Box 1), primary
steps of raw data processing include aligning and counting reads or
molecular barcodes per genomic region of interest, with sequenc-
ing barcodes defining the spatial sector. Image-based methods, by
contrast, utilize fluorescent signals, often measured across multiple
fluorescence channels and cycles. These signals must be detected,
quantified and decoded to their corresponding target molecule,
which can be defined by a specific combination
of colours across cycles. For whole-slide imaging, Integration
fields of view need to be stitched together into
one tissue image258. Background fluorescence,
be picked up by various topic and graph-based models. Mechanistic
models attempt to explain expression plasticity mediated by cell–cell
interactions and signalling at the cell interface107.
Spatial lineage tracing
Spatial RNA or DNA sequencing technologies provide wide but low
genomic coverage, which limits lineage tracing to copy number alter­
ations and a few point mutations37,279,280. Current protocols also have
a relatively low spatial resolution (Box 1). Microscopy methods target
specific point mutations and require prior knowledge of the clonal struc-
ture. Yet, they achieve a higher spatial resolution to map a more fine-
grained clonal structure across wide fields of view43. The incremental
nature of branching evolution also provides opportunities to associate
the genetic changes on each branch with phenotypic changes.
H&E slide Spatial analysis
Cells, morphology
micro-anatomy

Time
blur and signal crowding can negatively affect
the ability to decode signals243,259–262. Technolo-
gies with a super-cellular resolution require cell Space
decomposition algorithms, where known cell types
usually obtained from the single-cell data serve Genes with Genes with
Subclone-specific
as a reference43,129,263,264. Signals with a subcellular histology and significantly significantly
resolution are spatially aggregated either by clus- micro-anatomy higher expression higher expression
Genomics in red subclone in green subclone
tering265–268 or based on cell segmentation derived Mutations and
from nuclear DAPI signals, membrane markers or subclones
–log (P value)

haematoxylin and eosin (H&E) staining235,266,269,270.

Data integration
Single-cell resolved data can be mapped271 to
deconvolve larger spots or to annotate known cell 0
types. Single-cell whole transcriptomes or multi- Subclone-specific Log fold change
expression and
omics can be further mapped into space, leverag- tumour
Stromal cell type

ing a smaller common set of reference genes that microenvironment

Transcriptomics
are interrogated spatially270,272. Another option is
abundance

Expression
to integrate multiple spatially resolved modalities
from consecutive tissue sections. Alignment of
distorted sections and 3D reconstruction of a tissue
block can be done using image registration algo- Red Green
rithms273. Multimodal characterizations of micro- subclone subclone
scopically matched tissue areas, including spatial Cell neighbourhoods
transcriptomics, proteomics and histopathology, Subclone-specific Cellular neighbourhood
are also possible43,274. proteins
Proteomics
Spatial statistics Protein levels
Spatial differential expression models enable the
description of novel cell states showing distinct
spatial patterns that are difficult to recover from
single-cell data275,276. A group of cells may appear
enriched in a recurrent way forming distinct
neighbourhoods137,139,277,278, a pattern which could Tumour lesion

Nature Reviews Genetics | Volume 24 | May 2023 | 295–313 302

Review article

Glossary

Adjuvant therapies Immune evasion Mutation Subclones

Cancer treatments given after primary
therapy (particularly following surgical
resection of a tumour) to reduce the risk
of relapse.
Carcinomas in situ
Neoplastic expansions of epithelial cells
that are confined to the normal tissue
structure in which they arose, without
invasion of the adjacent stroma.
Clones
A clone is a population of cells that
derive from a common ancestor.
Cancers are found to be clones deriving
from a single cell that expanded during
the lifetime of a host. Clonal alterations
are mutations that are present in all cells
of a cancer because their occurrence
preceded the expansion of the tumour.
Immune-checkpoint blockade
A type of anticancer immunotherapy
that blocks checkpoint proteins on
T cells or their targets on cancer cells
and promotes an immune response and
killing of the tumour.
The process by which tumour cells
develop several mechanisms that
help them to continue to grow and
expand by escaping immune control.
Immune evasion is thought to be
typically preceded by two other phases:
elimination, when the immune system
recognizes and eliminates tumour cells,
and equilibrium, when the pressure
from the immune system stalls tumour
growth and expansion, and negative
selection takes place.
Invasion
The process of penetration and spread
of cancer cells into the neighbouring
normal tissue. In the case of carcinomas,
invasion involves a breach of the
basement membrane.
Metastasis
A process of cancer spreading from the
primary disease site to lymph nodes or
other organs, resulting in the formation
of metastatic deposits known as
metastases.
Most recent common
ancestor
In cancer evolution, the most recent
founder cell from which all other
tumour cells have directly descended.
Changes in the DNA sequence that
are inherited across cell generations.
Sources of mutation are erroneous
replication, biochemical alterations of
DNA and failed DNA repair. To date, all
cells within the human body are found
to accumulate mutations over their
lifetime.
Negative selection
The removal of deleterious variants from
the population.
Neoantigen
A tumour-specific antigen that is the
result of a somatic coding mutation
in the corresponding part of a gene.
Antigens are protein fragments
presented on the cell surface by
the human leukocyte antigen (HLA)
complex and recognized by different
cells of the immune system.
Positive selection
The spread of advantageous alleles
within a population.
Selection
In evolution, natural selection denotes
the process of survival and reproduction
of the fittest organism within a given
environment.
Further clones emerging within a
tumour from one founder clone.
Subclonal mutations are limited to
a fraction of cancer cells and occur
during tumour expansion.
Tissue architecture
The micro-anatomical spatial
organization of the tissue. Typical
examples are layered epithelial tissues,
glands and crypts. Tissue architecture
can be combined with other means
of tissue organization such as
differentiation hierarchies.
Tumour ecosystems
The collective set of heterogeneous
cells in the vicinity of a tumour
comprising cancer cells and the TME.
Tumour microenvironment
(TME). A combination of non-tumour
cells, such as stromal and immune
cells, vessels, metabolites, signalling
molecules, and other extracellular
components among which tumour
cells exist.

high-resolution three-dimensional optical deconvolution was used to
study the physical features that might be associated with cellular inter-
actions139. In melanoma tissues, probable interactions were identified
amongst approximately 20% of immune cells at the tumour–stroma
interface (Fig. 3d). These interactions were often complex, involving
three or more cells (for example, a melanoma cell contacting two CD8+
cytotoxic T lymphocytes and one CD4+ Treg cell) with associated cell sur-
face molecular polarization and receptor–ligand interaction. In some
cases, cellular processes from macrophages extended more than 10 μm
to make contact with other cells. Standard fine sections would miss
many of these interactions, whereas basic proximity measurements
would tend to overestimate the nature of interactions. As discussed
in Box 2, although these analyses remain technically challenging and
have low throughput, they demonstrate the unique potential afforded
by spatial omics for the mapping of functional cellular interactions in
tissue space.
Tumour histology and macrostructure
The histological appearance of tumour cells and their micro-anatomical
structures also reflect the underlying genomic alterations as revealed
by two recent pan-cancer analyses using artificial intelligence-based
digital histopathology143,144. In some cases, histological patterns have
been found to be indicative even of distinct subclonal alterations43,145,146.
It is well established that tumours exhibit characteristic histological
structures indicative of their natural history (Fig. 3e). Many cancers
exhibit adjacent areas of carcinoma in situ, which are considered to be
precursor lesions. This phenomenon is well established in melanoma139
and many adenocarcinomas: in breast cancer up to two-thirds of cases
have histologically distinct areas of ductal carcinoma in situ147, around
15% of colorectal adenocarcinomas exhibit adjacent adenomas148, and
~45% of oesophageal adenocarcinomas show detectable areas of its
precursor lesion Barrett oesophagus149. Interestingly, the phylogenetic
and histological relationships between precursor lesions and invasive
cancer are not always aligned with patterns indicative of branching150,151
or multiclonal invasion21. Therefore, spatial genomic and transcrip-
tomic analyses using histologically informed dissection allow for the
exploration of mechanisms and phenotypes of disease progression in
greater detail, revealing subclone and histology-specific changes43.
At the scale of millimetres, cancers are often described in terms of a
tumour core that is separated from the surrounding stroma by an inva-
sive margin (Fig. 3e). Transcriptional diversity has been reported across
these landscapes128, with hypoxia and EMT-like signatures localizing

Nature Reviews Genetics | Volume 24 | May 2023 | 295–313 303

Review article
a Cytolysis of tumour Antitumour immunity ECM Single-cell data provide high molecular
cells by NK or CTL Oxygen inhibition by Treg cells detail of TME components but lack
spatial context
ECM remodelling by
CAFs facilitates tumour
cell migration

Tumour-promoting M2
macrophages are recruited
to hypoxic areas and release
Immune activation immuno-suppressive cytokines
by M1 macrophages
via cytokine release Capillary
Single-cell RNA sequencing:
High-resolution TME composition

Tumour cells CAFs

Immune activation of Endothelial cells T cells
CD4+ cell by B cell Dendritic cells B cells
Dendritic cell CD4+ T cell Macrophages
Antigen presentation to neighbouring T cells

Tumour inhibiting Tumour promoting

b Co-localization c Cellular neighbourhoods and structures

Cell type A Cell type B
Suppressed
expansion structure

Putative
interaction Blood
vessel
% % Vascular
0.3 0.4 stroma
Tumour lesion

Visium spots 0 0
B cell
aggregates

Tertiary lymphoid
Diffuse structure
d Interfaces T cells

Ligand X
Receptor Y Denritic
cell
DNA
Immunofluorescence reveals receptor–ligand
interactions such as PD1–PDL1 B cell T cell

e Histology Pathological annotation Clonal map

Carcinoma Invasive
in situ front with
EMT
Invasive Immune
cancer infiltration Immune-
with large excluded Genetically
cells area distinct
Tumour
lesion subclones
Hypoxic
Invasive area with
cancer necrosis
with small
cells Stroma

Nature Reviews Genetics | Volume 24 | May 2023 | 295–313 304

Review article
Fig. 3 | Mapping the cancer ecosystem. a, Cellular composition of the tumour
microenvironment (TME). The TME is composed of different cell types, involved
in tumour-suppressing (on the left) and tumour-promoting (on the right)
interactions with cancer cells. While TME composition can be captured by single-
cell omic approaches (such as single-cell RNA sequencing), its spatial context is
lost as a consequence of tissue dissociation. b, Cellular interactions. Spatial
co-localization and concomitant expression of receptors and ligands are
indications of cellular interactions. c, Cellular neighbourhoods. Spatially resolved
to the core and invasive margin, respectively101,152. The core–stroma
distinction is fundamental to common classifications of immune infil-
trates that correlate with cancer subtype, clinical outcome and therapy
response153–155. At one end of the immune infiltrate continuum are
inflamed tumours, where plentiful immune cells infiltrate the tumour
core, at the other end are immune deserts that are devoid of infiltrates,
and between these are immune-excluded tumours where immune cells
are largely restricted to the neighbouring stroma156. However, recent
spatial studies in pancreatic and breast cancer reveal that these immune
characteristics can be highly localized and coexist in different parts
of the same tumour137,157,158. A challenge will therefore be to decipher
how localized variation in immune response shapes the evolutionary
outcome of the tumour.
Evolution of the cancer ecosystem
The cancer ecosystem detailed in the previous section changes the
dynamics of cancer evolution as tumour cells face selection pressures
that maximize growth-promoting and minimize growth-inhibiting
interactions. Foremost are cell–cell and other local interactions with
the immune system but also with fibroblasts, vasculature and even
neurons, which can lead to the selection of specific genomic alterations
(Fig. 4a). The fact that the TME is itself spatially heterogeneous suggests
that tumour cells experience localized variations in selective pressure.
Evolution of immune evasion
Various parts of the immune system constrain tumour growth. The
observed range of associations between specific genetic alterations
and the immune system cells of the TME159,160 illustrates the presence of
selective pressures on cancer evolution. Neoantigens, which derive from
coding mutations in cancer genomes, have been reported to elicit T cell
reactivity in lung cancer161. In tumours with greater cytolytic activity,
loss of neoantigen presentation capabilities was more frequent, which is
indicative of greater selective pressures159,162. Other signs of selection for
immune evasion include mutations in CASP8, which serve as a potential
way to resist apoptosis induced by FASL159. Furthermore, gains of immu-
nosuppressors, such as CD274 (also known as PDL1), are widespread
among Hodgkin lymphomas and are detected in ~0.7% of solid tumours163.
The signs of immune evasion increase during later stages in cancer
evolution. No evidence of negative selection was reported in diploid
genomes of normal tissues and among clonal alterations in primary
cancers, indicating that there is no depletion of coding mutations
during normal evolution8,164. Yet, the signs of selection for immune
evasion become stronger after clonal expansion at the subclonal level.
For example, immune evasion via deletions of human leukocyte antigen
(HLA) class I alleles, predominantly subclonally and in metastases, has
been described in lung cancer165 and multiple other cancer types33
(Fig. 4a). This mode of evolution is plausible as it may require a certain
tumour size for the immune system to recognize its neoantigens; at this
stage, however, a mutation enabling escape from these new selective
pressures arises in a single cell of the tumour and seeds the growth
of a tumour subclone. While it is possible for it to replace all other
tumour cells, the timing at which this occurs depends on the strength
of immune control and its escape rate.
Nature Reviews Genetics | Volume 24 | May 2023 | 295–313
proteomic studies detect clinically relevant neighbourhoods of molecularly
related cell types137. d, Cellular interfaces. Subcellular resolution imaging reveals
receptor–ligand localization at cellular interfaces139. e, Tumour macrostructure.
Different tumour regions can be identified in some tumours based on histological
staining. Such histopathological findings can be supplemented and underpinned
by detailed molecular maps. CAFs, cancer-associated fibroblasts; CTL, cytotoxic
T lymphocyte; ECM, extracellular matrix; EMT, epithelial-to-mesenchymal
transition; NK, natural killer; Treg, regulatory T cell.
Furthermore, selective pressures of the immune system determine
prognosis and immune evasion is found in relapses. It has been reported
that strongly immunogenic neoantigens are prognostically favour-
able in pancreatic cancer166 but are preferentially lost at recurrence167.
In melanoma, mutations of JAK1 or JAK2, which inhibit cytolysis, and in
B2M, which reduce antigen presentation, are often found in samples
that acquired resistance to immune-checkpoint inhibitors168,169.
Maps of cancer–TME interactions
Establishing associations between genomic alterations and the micro-
environment can be challenging due to the large genomic heterogene-
ity between primary cancers. Conceptually, spatial genomics offers new
ways of analysing such interactions at high spatial resolution and also
within the same tumour. For intratumour diversity, the observations
occur in an isogenic background and the set of genomic differences
between subclones is small compared to the large intertumour diver-
sity (Fig. 4b). A proof of concept has been established in breast cancer,
where various degrees of immune infiltration were found at the level
of cancer subclones43. Interestingly, such observations could also be
linked to the stage of invasion, with some TME regions defined by the
subclonal lineage and others mostly by histological progression from
carcinoma in situ to invasive cancer. Direct experimental evidence was
provided by Perturb-map, a mouse in vivo spatial CRISPR screen, which
revealed that specific gene knockouts lead to characteristic changes in
the immune microenvironment. For example, it was shown that T cell
infiltration is increased by Socs1 knockout in the tumour but decreased
by Tgfbr2 knockout170 (Fig. 4b).
Further cell types beyond immune cells have been associ-
ated with cancer evolution. Fibroblasts are an extremely versatile
group of cells that perform various roles that are exploited during
tumour growth. For example, cancer-associated fibroblasts can be
educated by cancer cells with gain-of-function mutations in TP53 to cre-
ate a pro-tumorigenic microenvironment171. Furthermore, fibroblasts
can amplify tumour-intrinsic oncogenic signalling in KRASG12D-mutant
pancreatic adenocarcinoma172. Spatial genomics adds to these obser-
vations. In breast cancer, imaging mass cytometry (Box 1) revealed
enrichment of different fibroblast and myofibroblast populations
in TP53-mutant and genomically unstable cancers173; however, the
nature of those associations is unclear. The aforementioned Perturb-
map screening revealed that Tgfbr2 knockout created a fibro-muci-
nous stroma characterized by TGFβ-activated fibroblasts and T cell
exclusion170.
Another cell-extrinsic and environmental factor influencing can-
cer evolution is hypoxia, the often localized depletion of oxygen in
tumours; it is well established as a marker of poor outcomes174 and
has been shown to select for certain mutations such as TP53 (ref.175).
A recent study combining scRNA-seq and immunofluorescence found
that hypoxic microniches contain quiescent cancer cells resistant to
T cell attack176.
Lastly, spatial genomics may also shed light on how different can-
cer subclones interact with each other. Reports of clonal cooperation,
observed in heterologous xenotransplants, date back to at least 1989
(ref.177). Elements of clonal cooperation have since been described as
305
Review article

Recruits Cancer cell

Neoantigen

Fibroblast
Presents
antigen
HLA1 Lost HLA1
Cytolysis
Microenvironment 1 follows the expansion of the subclone

Cytotoxic
T cell

Microenvironment 2
creates selective pressure
and favours the subclone
that developed immune
evasion via loss of HLA1

b
Fibro-mucinous stroma
Tgfbr2 knockout
tumour

Fibroblasts

T cells

Socs1 knockout
tumour
Tumour cells Fibroblasts

Primary tumour

Seeding in
different organs

Metastastases

Favourable Unfavourable Unfavourable Favourable

microenvironment microenvironment microenvironment microenvironment

Lung cancer subclone survives in brain and Colon cancer subclone survives in liver and
gives rise to metastasis but does not gives rise to metastasis but does not
metastasize in liver metastasize in brain

Nature Reviews Genetics | Volume 24 | May 2023 | 295–313 306

Review article
Fig. 4 | Evolution of the tumour ecosystem. a, Co-evolution of tumour cells
and microenvironments. Some tumour subclones attract characteristic tumour
microenvironments (TMEs), which subsequently spatially overlap with the
clonal territories. Alternatively, microenvironments may select for specific
tumour characteristics such as immune evasion; here, immune evasion is shown
as impaired antigen presentation due to loss of expression of human leukocyte
antigen 1 (HLA1). b, Examples of experimentally observed associations between
involving direct and microenvironmentally mediated interactions178–182.
The ability to comprehensively map genomic evolution, subclone-
specific gene expression and TMEs is thus likely to further elucidate
the full extent and mechanisms of this phenomenon.
Disseminated tumour cells, metastatic niches and
organotropism
Metastases are seeded by disseminated tumour cells (DTCs). The
mechanisms that determine whether any given DTC survives, enters a
dormant state or forms a metastasis are poorly defined, but it is evident
that these cells can exploit a variety of existing cellular interactions and
spatial microenvironments. For example, dissemination was shown
to depend on environmental triggers tied to the circadian rhythm183.
In breast cancer, DTCs physically associated with neutrophils were
shown to exhibit a more aggressive phenotype184. Additionally, aggre-
gates or even fusions of DTCs and macrophages have been described,
further illustrating that metastatic dissemination may also rely on
a range of cellular interactions185.
The existence of organotropism, whereby certain cancer types
and specific molecular alterations exhibit distinct patterns of meta-
static organ involvement, is well established and is often considered
in terms of the classical seed and soil hypothesis suggested by Paget,
which posits that certain tissues provide host environments suscep-
tible to specific tumour types79,186 (Fig. 4c). Organotropism seems to
be mostly determined by the cell of origin of the primary tumour, the
ability of DTCs to interact with the metastatic host environment and
the anatomical proximity of certain target organs such as liver meta­
stases in colorectal cancer. In breast cancer, brain metastases were
shown to be facilitated by the interactions with glutamatergic neurons187,
and a similar pattern was also observed in primary brain cancers where
neoplastic cells form networks with normal astrocytes via synapses188.
The notion that both intrinsic and microenvironmental features
of cellular populations in the primary tumour influence metastatic
rates is further supported by lineage-tracing studies in mouse mod-
els189. However, a genomic element also emerges. A recent pan-cancer
analysis of 25,000 metastases of 10 cancer types revealed 57 genetic
associations of organotropism within the same primary tumour type94.
Although the presence of many genomic alterations was associated
with increased rates of metastasis in multiple organs, there were also
examples of genomic alterations associated with reduced burden, for
example, RBM10 mutations being less prevalent in brain metastases of
lung cancers. However, it is worth noting that these genetic patterns
appeared to be mostly specific to certain primary tumour sites rather
than to target sites.
A series of studies reported that tumours prime target tissues to
create pre-metastatic niches facilitating colonization and metastatic
outgrowth190. This occurs prior to or on the arrival of DTCs through
soluble factors and extracellular vesicles released by the primary
tumour that can alter the systemic immune response and induce local-
ized regions of vascular and stromal remodelling in target organs.
Nature Reviews Genetics | Volume 24 | May 2023 | 295–313
clones and TMEs. The yellow Tgfbr2 knockout tumour clone exhibits a fibroblast-
enriched stroma with no presence of T cells, whereas the green Socs1 knockout
clone has a stroma infiltrated with T cells170. c, Organotropism and the seed and
soil hypothesis of metastasis. Cancer cells from the primary tumour migrate
throughout the body via blood vessels, with some of them able to seed metastatic
lesions in a distant organ. The target TME determines the fitness of clones at the
site of metastasis favouring some tumour lineages over others.
Furthermore, DTCs can enter dormant niches that facilitate long-
term survival and even evasion of therapy. In prostate cancer, DTCs
can compete with haematopoietic stem cells for the occupation of
established endosteal niches191, whereas perivascular niches have been
observed in bone, lung and brain192,193. In melanoma, age-dependent
changes of fibroblasts have been reported to induce emergence from
dormancy in lung deposits194. Similarly, in a mouse model of melanoma,
metastases in lymph nodes were found to promote immune evasion
and facilitate further metastasis in distant organs195. As all of these
phenomena involve cellular and spatial interactions, spatial genomic,
transcriptomic and proteomic analyses are likely to provide further
insights into these processes.
Translational opportunities
As discussed in the preceding sections, spatial genomic methods gener-
ate rich insights into the cellular composition, organization and interac-
tions that shape the evolving cancer ecosystem. For any given cancer,
the genome bears the scars of its unique, past evolutionary journey;
however, clinical interventions are designed to shape the future disease
course (Fig. 5). A major hope for cancer care is therefore that evolution-
ary features might be leveraged to better predict the clinical trajectory.
Because tumour evolution is a spatial process, spatially resolved fea-
tures may provide additional means to predict outcomes. Furthermore,
a better understanding of the cellular interactions within the ecosys-
tem might identify potential therapeutic vulnerabilities that could be
exploited to control the evolutionary process127,129,131,196,197. Initial studies
hint towards the opportunities afforded by these approaches for novel
biomarker discovery137,154,197, in understanding the clinical relevance of
genetic subclonal structure37,43,198 and in identifying therapeutic vulner-
abilities through a deepened understanding of cellular interactions
in both human samples129,131,154 and preclinical animal models170.
Spatial biomarker discovery
Spatial biomarkers predict clinical outcomes by leveraging information
about cellular organization or intercellular relationships. In its most
simplistic form, this approach might measure the distribution of a
certain cell type in relation to another. This is the basic concept of com-
monly used, histomorphological scoring of tertiary lymphoid struc-
tures or tumour infiltrating lymphocytes (TILs) within the tumour or
stromal compartment (Fig. 3c). Stromal TILs are associated with a supe-
rior prognosis and response to chemotherapy or immune-checkpoint
inhibitors in triple-negative breast cancer, non-small-cell lung cancer
and melanomas199–203. However, in isolation, the reliability of TILs as
a biomarker is hampered by heterogeneity and irreproducibility203.
Recent studies that use highly multiplexed, in situ, proteomic-based
assays have started to dissect this complexity by characterizing TIL
subtypes, expression, cellular co-localization and neighbourhood pat-
terns, identifying prognostically meaningful TIL-related patterns197,154.
In general, head-to-head clinical trials are needed to confirm whether
multiplex biomarkers carry additional prognostic and predictive value
307
Review article
Map of subclones Microenvironment
Blood vessel
TILs
Cancer
subclones Necrotic core
TLS
Anatomic location Evolutionary history
Genetic subclones
ne
Driver mutation
lo
bc
Su
Fig. 5 | Translational opportunities. Subclone maps, tumour microenvironment
(TME) composition and histological context each provide clinically and
prognostically relevant features. Integrating these levels of information may
help identify ill-fated clonal lineages that may not be found using either
compared to single biomarkers such as PDL1 immunohistochemis-
try204. Notably, the Immunoscore205 is a digital pathology-based assay
that measures CD3+ and CD8+ lymphocyte density in the tumour core
and invasive margin and has been proven to surpass clinical staging
systems for disease-free survival prediction, demonstrating that
higher-dimension datasets could be clinically transformative206.
Beyond a role in refining histology-based biomarkers, spatial
omic approaches can also facilitate de novo biomarker discovery. This
was demonstrated by two breast cancer studies (discussed in the sec-
tion ‘Mapping the cancer ecosystem’) that extracted recurrent cellular
communities from multiplex proteomic and morphological data using
machine learning algorithms137,138. Another use of multiplexed tissue
imaging was an ovarian cancer study, which found that interactions
of exhausted CD8+ T cells, PDL1+ macrophages and PDL1+ tumour cells
were linked to treatment response207. In these studies, the power to
detect clinically meaningful features was derived from the fact that
hundreds of samples could be analysed at high throughput by using
tissue microarrays. However, the trade-off was that only a tiny part
of the main tumour mass was sampled (core needle biopsies with a
diameter of 0.6–0.8 mm), whereas the invasive margin and stroma
were largely unsampled. An important next step for spatial mole­
cular analyses is to derive catalogues of typical cancer ecosystems
at a tissue-wide scale and at single-cell resolution. These reference
sets could serve as a basis for designing more focused studies that
might take advantage of lower-plex, higher-throughput or rational
subsampling approaches. The scalability and reproducibility of omic
Nature Reviews Genetics | Volume 24 | May 2023 | 295–313
Histological context
Normal
Invasive cancer
Carcinoma
in situ
Basement
membrane
Present Future
Genetics, histology
and TME suggest
high risk of
metastasis if left
untreated
Metastasis
y
og
TM
ol
st
Hi
component alone. For example, the metastatic potential of two invasive
subclonal lineages may differ based not only on the portfolio of their driver
mutations but also on the TMEs that they inhabit. TILs, tumour infiltrating
lymphocytes; TLS, tertiary lymphoid structure.
data will be critical factors for successful implementation within
large-scale clinical studies.
Clinically relevant subclones
An unanswered question for personalized medicine is how to deal
with subclonal driver mutations that can be targeted by antitumour
therapy. By integrating additional layers of spatial data, it is possi-
ble to establish the spatial context and thus the clinical relevance of
subclonal alterations43,208 (Fig. 5). For example, systemic therapies
administered after primary breast cancer surgery are directed towards
covert metastatic disease, with the intention of eliminating these cells
before metastatic deposits have a chance to form. As we cannot directly
assay disseminated tumour cells, treatment paradigms are based upon
the properties of the primary tumour. It is reasonable to assume that a
subclonal mutation within the invasive cancer might also be carried by
disseminated tumour cells and could drive metastasis formation.
By contrast, if the mutation is entirely restricted to precursor lesions,
and hence a clone that has not yet even exhibited invasive capacity,
we can be relatively confident that it will not be a driver of metastasis.
Integrating spatial transcriptomic or proteomic data provide
further insights into the TME and the gene expression properties of
subclones. This additional characterization will be helpful for deriving
predictions of the functional relevance of a particular subclonal driver
mutation, such as by considering a range of intrinsic and extrinsic
phenotypes, including vascular invasion or a high-risk TME niche
(Fig. 5). Although further research is needed, it is possible that spatial
308
Review article
analyses could be used in early detection settings or to complement
histopathology-determined completeness of excision, with the latter
application yielding opportunities for biomarker-driven surgical and
adjuvant radiotherapy clinical trials.
Preclinical studies
Although spatial genomic analysis enables reconstruction of the evo-
lutionary past and prediction of the clinical future, spatial genomic
experiments in animal models can add direct mechanistic evidence
of the roles of particular cancer-associated mutations and/or the
effects of spatial aspects of the cancer TME. Such experiments are
expected to provide clinically important insights into the extracellular
consequences of perturbations in cell-intrinsic processes, with many
opportunities for biomarker discovery196,209. A particularly exciting
approach is the combination of spatial profiling with in vivo CRISPR
screens131,170. Using this approach, both cell-intrinsic phenotypic and
local TME associations of tens of genetically distinct clones can be
studied simultaneously in a single animal, leading to insights into
the precise mechanisms that lead to diverse clinical fates of geneti-
cally divergent clones that nonetheless share many genetic and host
characteristics170. These models could lead the way in developing pan-
ecosystem therapeutic strategies that simultaneously target cell-
intrinsic vulnerabilities and harness the properties of the TME in an
attempt to terminate cancer evolution.
Conclusions and future perspectives
While it is widely accepted that the population genetics of carcinogen-
esis can be well described in terms of somatic evolution, the spatiotem-
poral details of the process are not fully understood. Open questions
remain over the extent to which cellular interactions with the TME
and micro-anatomical constraints influence cancer evolution. This
situation begins to change with the availability of spatial genomic,
transcriptomic and proteomic technologies that enable charting of
the growth patterns of distinct subclones and characterization of the
composition and structure of their microenvironments. Connecting
these different levels of intratumour heterogeneity is therefore key to
understanding how tumours grow as it reveals how tumour subclones
interact with their microenvironment and how this may lead to the
selection of aggressive traits.
Although there is a broad range of technologies with different
advantages and disadvantages, it emerges that cellular resolution
is essential for understanding the micro-anatomy of the cancer eco-
system and localizing molecular signals to specific cells. Mapping
how cells are spatially organized into subclones and neighbourhoods
and how different cells interface is a key step for understanding the
transcriptional diversity revealed by catalogues of scRNA-seq studies.
Another consideration is the field of view. A typical primary tumour is
more than 1 cm in diameter at diagnosis; hence, technologies enabling
the mapping of entire tumour sections are desirable. Lastly, enabling
spatially resolved genomics with the resolution of single nucleotides
is important for mapping the subclonal landscape. Whereas spatial
transcriptomic and proteomic approaches are reaching maturity and
are available as commercial platforms, spatial genomics lags behind.
A reason for this is the limited amount of DNA per cell, compared to
the much greater number of copies of RNA and protein, which may be
compounded by tissue sections only containing fractions of a nucleus.
Even though it is possible to calculate copy number alterations using
low-coverage DNA sequencing or even RNA sequencing, it remains
challenging to estimate phylogenetic relationships and distances.
Nature Reviews Genetics | Volume 24 | May 2023 | 295–313
Developing computational approaches such as integrating mathemati-
cal modelling210, mathematical oncology and artificial intelligence-
based algorithms may help to overcome these challenges and generate
further quantitative insights.
These challenges notwithstanding, the opportunities for spatial
cancer genomics are formidable. Charting the tumour ecosystem
with clonal resolution as well as cellular and molecular details of the
TME will have great value. Doing so across whole tumour sections and
cancer types will create insightful atlases that not only provide snap-
shots of cancer composition but also insights into past tumour evolu-
tion. In addition to a more detailed understanding of tumour growth,
these atlases have the potential to reveal the mechanism by which the
microenvironment is co-opted by tumour cells and how the immune
system suppresses tumour proliferation. These processes are already
exploited by immune-checkpoint inhibitors, and enhanced molecular
and spatial detail may yield further therapeutic targets or facilitate
more effective use of existing therapies. Furthermore, the derivation
of spatial biomarkers and the detection of diagnostic applications of
spatial genomics are likely to inform treatment; this is especially
the case now that spatial genomic, transcriptomic, proteomic and
metabolomic methods are starting to be used in clinical trials. Such a
multimodal approach is required to elucidate the inherent complexity
of the TME — the result will be a revolution of histopathology that is
genomically and molecularly informed.
The availability of various single-cell and spatially resolved assays
and the emerging insights that these approaches offer thus warrant
new concerted efforts. Previous initiatives, such as The Cancer Genome
Atlas211 and the International Cancer Genome Consortium212,213, have
created valuable community resources detailing the molecular and
genomic properties across cancers. It thus seems natural for emerging
projects, such as The Human Tumour Atlas Network214, to follow in this
spirit in order to create the next generation of molecularly resolved
cancer atlases.
Published online: 9 December 2022
References
1. Cairns, J. Mutation selection and the natural history of cancer. Nature 255, 197–200 (1975).
A seminal paper discussing the role of somatic evolution as well as tissue anatomies
protective against evolution.
2. Nowell, P. C. The clonal evolution of tumor cell populations. Science 194, 23–28 (1976).
3. Martincorena, I. & Campbell, P. J. Somatic mutation in cancer and normal cells. Science
349, 1483–1489 (2015).
4. Moore, L. et al. The mutational landscape of human somatic and germline cells. Nature
597, 381–386 (2021).
A pan-body atlas of mutations and clonal expansion in normal tissues.
5. Li, R. et al. A body map of somatic mutagenesis in morphologically normal human
tissues. Nature 597, 398–403 (2021).
6. Alexandrov, L. B. et al. The repertoire of mutational signatures in human cancer. Nature
578, 94–101 (2020).
7. Greenman, C., Wooster, R., Futreal, P. A., Stratton, M. R. & Easton, D. F. Statistical analysis
of pathogenicity of somatic mutations in cancer. Genetics 173, 2187–2198 (2006).
8. Martincorena, I. et al. Universal patterns of selection in cancer and somatic tissues. Cell
171, 1029–1041.e21 (2017).
9. Lawrence, M. S. et al. Mutational heterogeneity in cancer and the search for new
cancer-associated genes. Nature 499, 214–218 (2013).
10. Gonzalez-Perez, A. et al. IntOGen-mutations identifies cancer drivers across tumor types.
Nat. Methods 10, 1081–1082 (2013).
11. Sondka, Z. et al. The COSMIC Cancer Gene Census: describing genetic dysfunction
across all human cancers. Nat. Rev. Cancer 18, 696–705 (2018).
12. Hanahan, D. & Weinberg, R. A. The hallmarks of cancer. Cell 100, 57–70 (2000).
13. Hanahan, D. Hallmarks of cancer: new dimensions. Cancer Discov. 12, 31–46 (2022).
14. Shah, S. P. et al. Mutational evolution in a lobular breast tumour profiled at single
nucleotide resolution. Nature 461, 809–813 (2009).
15. Nik-Zainal, S. et al. The life history of 21 breast cancers. Cell 149, 994–1007 (2012).
A seminal paper demonstrating that subclonal diversity is common in breast cancer
and reconstructing the evolutionary history of copy number alterations.
309
Review article
16. McGranahan, N. et al. Clonal status of actionable driver events and the timing
of mutational processes in cancer evolution. Sci. Transl. Med. 7, 283ra54 (2015).
17. Andor, N. et al. Pan-cancer analysis of the extent and consequences of intratumor
heterogeneity. Nat. Med. 22, 105–113 (2015).
18. Dentro, S. C. et al. Characterizing genetic intra-tumor heterogeneity across 2658 human
cancer genomes. Cell 184, 2239–2254.e39 (2021).
A demonstration of universal subclonal divergence across cancer types.
19. Navin, N. et al. Tumour evolution inferred by single-cell sequencing. Nature 472, 90–94
(2011).
A study pioneering the use of single-cell sequencing for understanding cancer
evolution.
20. Wang, Y. et al. Clonal evolution in breast cancer revealed by single nucleus genome
sequencing. Nature 512, 155–160 (2014).
21. Casasent, A. K. et al. Multiclonal invasion in breast tumors identified by topographic
single. Cell Sequencing. Cell 172, 205–217.e12 (2018).
22. McPherson, A. et al. Divergent modes of clonal spread and intraperitoneal mixing
in high-grade serous ovarian cancer. Nat. Genet. 48, 758–767 (2016).
23. Laks, E. et al. Clonal decomposition and DNA replication states defined by scaled
single-cell genome sequencing. Cell 179, 1207–1221.e22 (2019).
24. Salehi, S. et al. Clonal fitness inferred from time-series modelling of single-cell cancer
genomes. Nature 595, 585–590 (2021).
25. Williams, M. J., Werner, B., Barnes, C. P., Graham, T. A. & Sottoriva, A. Identification
of neutral tumor evolution across cancer types. Nat. Genet. 48, 238–244 (2016).
26. Navin, N. et al. Inferring tumor progression from genomic heterogeneity. Genome Res.
20, 68–80 (2010).
27. Gerlinger, M. et al. Intratumor heterogeneity and branched evolution revealed
by multiregion sequencing. N. Engl. J. Med. 366, 883–892 (2012).
One of the first studies using whole-genome sequencing to study patterns
of branching and subclonal variegation in kidney cancer.
28. de Bruin, E. C. et al. Spatial and temporal diversity in genomic instability processes
defines lung cancer evolution. Science 346, 251–256 (2014).
29. Gerlinger, M. et al. Genomic architecture and evolution of clear cell renal cell
carcinomas defined by multiregion sequencing. Nat. Genet. 46, 225–233 (2014).
30. Yates, L. R. et al. Subclonal diversification of primary breast cancer revealed by
multiregion sequencing. Nat. Med. 21, 751–759 (2015).
31. Morrissy, A. S. et al. Spatial heterogeneity in medulloblastoma. Nat. Genet. 49, 780–788
(2017).
32. Jamal-Hanjani, M. et al. Tracking the evolution of non-small-cell lung cancer. N. Engl.
J. Med. 376, 2109–2121 (2017).
33. Watkins, T. B. K. et al. Pervasive chromosomal instability and karyotype order in tumour
evolution. Nature 587, 126–132 (2020).
34. Heide, T. et al. The co-evolution of the genome and epigenome in colorectal cancer.
Nature 611, 733–743 (2022).
35. Woodcock, D. J. et al. Prostate cancer evolution from multilineage primary to single
lineage metastases with implications for liquid biopsy. Nat. Commun. 11, 5070 (2020).
36. Grossmann, S. et al. Development, maturation, and maintenance of human prostate
inferred from somatic mutations. Cell Stem Cell 28, 1262–1274.e5 (2021).
37. Zhao, T. et al. Spatial genomics enables multi-modal study of clonal heterogeneity
in tissues. Nature 601, 85–91 (2022).
38. Heide, T. et al. Multiregion human bladder cancer sequencing reveals tumour evolution,
bladder cancer phenotypes and implications for targeted therapy. J. Pathol. 248,
230–242 (2019).
39. Su, F. et al. Spatial intratumor genomic heterogeneity within localized prostate cancer
revealed by single-nucleus sequencing. Eur. Urol. 74, 551–559 (2018).
40. Bao, L. et al. Coexisting genomic aberrations associated with lymph node metastasis
in breast cancer. J. Clin. Invest. 128, 2310–2324 (2018).
41. Househam, J. et al. Phenotypic plasticity and genetic control in colorectal cancer
evolution. Nature 611, 744–753 (2022).
42. Biswas, D. et al. A clonal expression biomarker associates with lung cancer mortality.
Nat. Med. 25, 1540–1548 (2019).
43. Lomakin, A., Svedlund, J., Strell, C., Gataric, M. & Shmatko, A. Spatial genomics maps
the structure, character and evolution of cancer clones. Nature 611, 594–602 (2022).
44. Sottoriva, A. et al. A Big Bang model of human colorectal tumor growth. Nat. Genet. 47,
209–216 (2015).
45. Waclaw, B. et al. A spatial model predicts that dispersal and cell turnover limit intratumour
heterogeneity. Nature 525, 261–264 (2015).
46. Gallaher, J. A., Brown, J. S. & Anderson, A. R. A. The impact of proliferation-migration
tradeoffs on phenotypic evolution in cancer. Sci. Rep. 9, 2425 (2019).
47. Lewis, S. M. et al. Spatial omics and multiplexed imaging to explore cancer biology.
Nat. Methods 18, 997–1012 (2021).
48. Lieberman, E., Hauert, C. & Nowak, M. A. Evolutionary dynamics on graphs. Nature 433,
312–316 (2005).
49. Frean, M., Rainey, P. B. & Traulsen, A. The effect of population structure on the rate
of evolution. Proc. Biol. Sci. 280, 20130211 (2013).
50. Tkadlec, J., Pavlogiannis, A., Chatterjee, K. & Nowak, M. A. Population structure
determines the tradeoff between fixation probability and fixation time. Commun. Biol. 2,
138 (2019).
51. Lopes, J. V., Pacheco, J. M. & Dingli, D. Acquired hematopoietic stem-cell disorders
and mammalian size. Blood 110, 4120–4122 (2007).
Nature Reviews Genetics | Volume 24 | May 2023 | 295–313
52. Lee-Six, H. et al. Population dynamics of normal human blood inferred from somatic
mutations. Nature 561, 473–478 (2018).
53. Sender, R., Fuchs, S. & Milo, R. Revised estimates for the number of human and bacteria
cells in the body. PLoS Biol. 14, e1002533 (2016).
54. Humphries, A. & Wright, N. A. Colonic crypt organization and tumorigenesis. Nat. Rev.
Cancer 8, 415–424 (2008).
55. Nowak, M. A., Michor, F. & Iwasa, Y. The linear process of somatic evolution. Proc. Natl
Acad. Sci. USA 100, 14966–14969 (2003).
56. Sender, R. & Milo, R. The distribution of cellular turnover in the human body. Nat. Med.
27, 45–48 (2021).
57. Lee-Six, H. et al. The landscape of somatic mutation in normal colorectal epithelial cells.
Nature 574, 532–537 (2019).
58. Moore, L. et al. The mutational landscape of normal human endometrial epithelium.
Nature 580, 640–646 (2020).
59. Genovese, G. et al. Clonal hematopoiesis and blood-cancer risk inferred from blood
DNA sequence. N. Engl. J. Med. 371, 2477–2487 (2014).
60. Jaiswal, S. et al. Age-related clonal hematopoiesis associated with adverse outcomes.
N. Engl. J. Med. 371, 2488–2498 (2014).
Genovese et al.59 and Jaiswal et al.60 revealed the frequent presence of macroscopic
clones with leukaemogenic mutations in the blood of healthy individuals.
61. Martincorena, I. et al. Tumor evolution. High burden and pervasive positive selection
of somatic mutations in normal human skin. Science 348, 880–886 (2015).
A pioneering study revealing that normal skin is a patchwork of microscopic mutant
clones.
62. Martincorena, I. et al. Somatic mutant clones colonize the human esophagus with age.
Science 362, 911–917 (2018).
63. Lawson, A. R. J. et al. Extensive heterogeneity in somatic mutation and selection in the
human bladder. Science 370, 75–82 (2020).
64. Colom, B. et al. Mutant clones in normal epithelium outcompete and eliminate emerging
tumours. Nature 598, 510–514 (2021).
65. Fowler, J. C. et al. Selection of oncogenic mutant clones in normal human skin varies
with body site. Cancer Discov. 11, 340–361 (2021).
66. Abby, E. et al. Notch1 mutation drives clonal expansion in normal esophageal epithelium
but impairs tumor growth. Preprint at bioRxiv https://doi.org/10.1101/2021.06.18.448956
(2021).
67. Louis, D. N. et al. The 2007 WHO classification of tumours of the central nervous system.
Acta Neuropathol. 114, 97–109 (2007).
68. Elston, C. W. & Ellis, I. O. Pathological prognostic factors in breast cancer. I. The value
of histological grade in breast cancer: experience from a large study with long-term
follow-up. Histopathology 19, 403–410 (1991).
69. Epstein, J. I. et al. The 2014 international society of urological pathology (ISUP) consensus
conference on gleason grading of prostatic carcinoma: definition of grading patterns
and proposal for a new grading system. Am. J. Surg. Pathol. 40, 244–252 (2016).
70. Greenough, R. B. Varying degrees of malignancy in cancer of the breast. J. Cancer Res. 9,
453–463 (1925).
71. Polyak, K. & Weinberg, R. A. Transitions between epithelial and mesenchymal states:
acquisition of malignant and stem cell traits. Nat. Rev. Cancer 9, 265–273 (2009).
72. Morin, P. J. et al. Activation of β-Catenin-Tcf signaling in colon cancer by mutations
in β-catenin or APC. Science 275, 1787–1790 (1997).
73. West, J., Schenck, R. O., Gatenbee, C., Robertson-Tessi, M. & Anderson, A. R. A. Normal
tissue architecture determines the evolutionary course of cancer. Nat. Commun. 12,
2060 (2021).
74. Noble, R. et al. Spatial structure governs the mode of tumour evolution. Nat. Ecol. Evol. 6,
207–217 (2022).
75. Fearon, E. R. & Vogelstein, B. A genetic model for colorectal tumorigenesis. Cell 61,
759–767 (1990).
76. Gerstung, M. et al. The evolutionary history of 2658 cancers. Nature 578, 122–128 (2020).
77. Baslan, T. et al. Ordered and deterministic cancer genome evolution after p53 loss.
Nature 608, 795–802 (2022).
78. Lambert, A. W., Pattabiraman, D. R. & Weinberg, R. A. Emerging biological principles
of metastasis. Cell 168, 670–691 (2017).
79. Fidler, I. J. The pathogenesis of cancer metastasis: the ‘seed and soil’ hypothesis
revisited. Nat. Rev. Cancer 3, 453–458 (2003).
80. Jones, S. et al. Comparative lesion sequencing provides insights into tumor evolution.
Proc. Natl Acad. Sci. USA 105, 4283–4288 (2008).
81. Mamlouk, S. et al. DNA copy number changes define spatial patterns of heterogeneity
in colorectal cancer. Nat. Commun. 8, 14093 (2017).
82. Zhou, H. et al. Multi-region exome sequencing reveals the intratumoral heterogeneity
of surgically resected small cell lung cancer. Nat. Commun. 12, 5431 (2021).
83. Yates, L. R. et al. Genomic evolution of breast cancer metastasis and relapse. Cancer Cell
32, 169–184.e7 (2017).
84. Noorani, A. et al. Genomic evidence supports a clonal diaspora model for metastases
of esophageal adenocarcinoma. Nat. Genet. 52, 74–83 (2020).
85. Karlsson, J. et al. Four evolutionary trajectories underlie genetic intratumoral variation
in childhood cancer. Nat. Genet. 50, 944–950 (2018).
86. Gundem, G. et al. The evolutionary history of lethal metastatic prostate cancer. Nature
520, 353–357 (2015).
87. Hu, Z., Li, Z., Ma, Z. & Curtis, C. Multi-cancer analysis of clonality and the timing of systemic
spread in paired primary tumors and metastases. Nat. Genet. 52, 701–708 (2020).
310
Review article
88. Brastianos, P. K. et al. Genomic characterization of brain metastases reveals branched
evolution and potential therapeutic targets. Cancer Discov. 5, 1164–1177 (2015).
89. Brown, D. et al. Phylogenetic analysis of metastatic progression in breast cancer using
somatic mutations and copy number aberrations. Nat. Commun. 8, 14944 (2017).
90. De Mattos-Arruda, L. et al. The genomic and immune landscapes of lethal metastatic
breast cancer. Cell Rep. 27, 2690–2708.e10 (2019).
91. Aceto, N. et al. Circulating tumor cell clusters are oligoclonal precursors of breast cancer
metastasis. Cell 158, 1110–1122 (2014).
92. Priestley, P. et al. Pan-cancer whole-genome analyses of metastatic solid tumours. Nature
575, 210–216 (2019).
93. Martínez-Jiménez, F. et al. Pan-cancer whole genome comparison of primary and
metastatic solid tumors. Preprint at bioRxiv https://doi.org/10.1101/2022.06.17.496528
(2022).
94. Nguyen, B. et al. Genomic characterization of metastatic patterns from prospective
clinical sequencing of 25,000 patients. Cell 185, 563–575.e11 (2022).
One of the largest genomic analyses of metastasis and organotropism to date.
95. Reiter, J. G. et al. An analysis of genetic heterogeneity in untreated cancers. Nat. Rev.
Cancer 19, 639–650 (2019).
96. Aran, D., Sirota, M. & Butte, A. J. Systematic pan-cancer analysis of tumour purity.
Nat. Commun. 6, 8971 (2015).
97. Somarelli, J. A. The hallmarks of cancer as ecologically driven phenotypes. Front. Ecol.
Evol. 9, 661583 (2021).
98. Darmanis, S. et al. Single-cell RNA-Seq analysis of infiltrating neoplastic cells at the
migrating front of human glioblastoma. Cell Rep. 21, 1399–1410 (2017).
99. Venteicher, A. S. et al. Decoupling genetics, lineages, and microenvironment
in IDH-mutant gliomas by single-cell RNA-seq. Science 355, 1391 (2017).
100. Pombo Antunes, A. R. et al. Single-cell profiling of myeloid cells in glioblastoma across
species and disease stage reveals macrophage competition and specialization.
Nat. Neurosci. 24, 595–610 (2021).
101. Puram, S. V. et al. Single-cell transcriptomic analysis of primary and metastatic tumor
ecosystems in head and neck cancer. Cell 171, 1611–1624.e24 (2017).
102. Tirosh, I. et al. Dissecting the multicellular ecosystem of metastatic melanoma by single-
cell RNA-seq. Science 352, 189–196 (2016).
103. Izar, B. et al. A single-cell landscape of high-grade serous ovarian cancer. Nat. Med. 26,
1271–1279 (2020).
104. Lambrechts, D. et al. Phenotype molding of stromal cells in the lung tumor
microenvironment. Nat. Med. 24, 1277–1289 (2018).
105. Chung, W. et al. Single-cell RNA-seq enables comprehensive tumour and immune cell
profiling in primary breast cancer. Nat. Commun. 8, 15081 (2017).
106. Schapiro, D. et al. histoCAT: analysis of cell phenotypes and interactions in multiplex
image cytometry data. Nat. Methods 14, 873–876 (2017).
107. Arnol, D., Schapiro, D., Bodenmiller, B., Saez-Rodriguez, J. & Stegle, O. Modeling cell-cell
interactions from spatial molecular data with spatial variance component analysis.
Cell Rep. 29, 202–211.e6 (2019).
108. Hanahan, D. & Weinberg, R. A. Hallmarks of cancer: the next generation. Cell 144,
646–674 (2011).
109. Raskov, H., Orhan, A., Christensen, J. P. & Gögenur, I. Cytotoxic CD8+ T cells in cancer
and cancer immunotherapy. Br. J. Cancer 124, 359–367 (2021).
110. Philip, M. & Schietinger, A. CD8+ T cell differentiation and dysfunction in cancer. Nat. Rev.
Immunol. 22, 209–223 (2022).
111. Borst, J., Ahrends, T., Bąbała, N., Melief, C. J. M. & Kastenmüller, W. CD4+ T cell help
in cancer immunology and immunotherapy. Nat. Rev. Immunol. 18, 635–647 (2018).
112. Waldman, A. D., Fritz, J. M. & Lenardo, M. J. A guide to cancer immunotherapy: from T cell
basic science to clinical practice. Nat. Rev. Immunol. 20, 651–668 (2020).
113. Sharonov, G. V., Serebrovskaya, E. O., Yuzhakova, D. V., Britanova, O. V. & Chudakov, D. M.
B cells, plasma cells and antibody repertoires in the tumour microenvironment. Nat. Rev.
Immunol. 20, 294–307 (2020).
114. Schumacher, T. N. & Thommen, D. S. Tertiary lymphoid structures in cancer. Science 375,
eabf9419 (2022).
115. Sautès-Fridman, C., Petitprez, F., Calderaro, J. & Fridman, W. H. Tertiary lymphoid
structures in the era of cancer immunotherapy. Nat. Rev. Cancer 19, 307–325 (2019).
116. Wolf, N. K., Kissiov, D. U. & Raulet, D. H. Roles of natural killer cells in immunity to
cancer, and applications to immunotherapy. Nat. Rev. Immunol. https://doi.org/10.1038/
s41577-022-00732-1 (2022).
117. Goswami, S., Anandhan, S., Raychaudhuri, D. & Sharma, P. Myeloid cell-targeted
therapies for solid tumours. Nat. Rev. Immunol. https://doi.org/10.1038/s41577-022-
00737-w (2022).
118. Murdoch, C., Muthana, M., Coffelt, S. B. & Lewis, C. E. The role of myeloid cells in the
promotion of tumour angiogenesis. Nat. Rev. Cancer 8, 618–631 (2008).
119. DeNardo, D. G. & Ruffell, B. Macrophages as regulators of tumour immunity and
immunotherapy. Nat. Rev. Immunol. 19, 369–382 (2019).
120. Valkenburg, K. C., de Groot, A. E. & Pienta, K. J. Targeting the tumour stroma to improve
cancer therapy. Nat. Rev. Clin. Oncol. 15, 366–381 (2018).
121. Kalluri, R. The biology and function of fibroblasts in cancer. Nat. Rev. Cancer 16, 582–598
(2016).
122. Lendahl, U., Muhl, L. & Betsholtz, C. Identification, discrimination and heterogeneity
of fibroblasts. Nat. Commun. 13, 3409 (2022).
123. Sahai, E. et al. A framework for advancing our understanding of cancer-associated
fibroblasts. Nat. Rev. Cancer 20, 174–186 (2020).
Nature Reviews Genetics | Volume 24 | May 2023 | 295–313
124. Davidson, S. et al. Single-cell RNA sequencing reveals a dynamic stromal niche that
supports tumor growth. Cell Rep. 31, 107628 (2020).
125. Ståhl, P. L. et al. Visualization and analysis of gene expression in tissue sections by spatial
transcriptomics. Science 353, 78–82 (2016).
126. Wu, S. Z. et al. A single-cell and spatially resolved atlas of human breast cancers.
Nat. Genet. 53, 1334–1347 (2021).
127. Andersson, A. et al. Spatial deconvolution of HER2-positive breast cancer delineates
tumor-associated cell type interactions. Nat. Commun. 12, 6012 (2021).
128. Berglund, E. et al. Spatial maps of prostate cancer transcriptomes reveal an unexplored
landscape of heterogeneity. Nat. Commun. 9, 2419 (2018).
129. Moncada, R. et al. Integrating microarray-based spatial transcriptomics and single-cell
RNA-seq reveals tissue architecture in pancreatic ductal adenocarcinomas. Nat. Biotechnol.
38, 333–342 (2020).
130. Qi, J. et al. Single-cell and spatial analysis reveal interaction of FAP+fibroblasts and
SPP1+macrophages in colorectal cancer. Nat. Commun. 13, 1742 (2022).
131. Ji, A. L. et al. Multimodal analysis of composition and spatial architecture in human
squamous cell carcinoma. Cell 182, 497–514.e22 (2020).
132. Ravi, V. M. et al. T-cell dysfunction in the glioblastoma microenvironment is mediated
by myeloid cells releasing interleukin-10. Nat. Commun. 13, 925 (2022).
133. Wu, R. et al. Comprehensive analysis of spatial architecture in primary liver cancer.
Sci. Adv. 7, eabg3750 (2021).
134. Gouin, K. H. III et al. An N-Cadherin 2 expressing epithelial cell subpopulation predicts
response to surgery, chemotherapy and immunotherapy in bladder cancer. Nat. Commun.
12, 4906 (2021).
135. Erickson, A. et al. Spatially resolved clonal copy number alterations in benign and
malignant tissue. Nature 608, 360–367 (2022).
136. Barkley, D. et al. Cancer cell states recur across tumor types and form specific
interactions with the tumor microenvironment. Nat. Genet. 54, 1192–1201 (2022).
137. Danenberg, E. et al. Breast tumor microenvironment structures are associated with
genomic features and clinical outcome. Nat. Genet. 54, 660–669 (2022).
A detailed spatial proteomic analysis of more than 500 breast cancers revealing
recurrent cellular neighbourhoods associated with clinical outcomes.
138. Jackson, H. W. et al. The single-cell pathology landscape of breast cancer. Nature 578,
615–620 (2020).
139. Nirmal, A. J. et al. The spatial landscape of progression and immunoediting in primary
melanoma at single-cell resolution. Cancer Discov. 12, 1518–1541 (2022).
A detailed multiplexed spatial proteomics analysis of melanoma revealing spatially
variegated TMEs and cellular interactions.
140. Schürch, C. M. et al. Coordinated cellular neighborhoods orchestrate antitumoral
immunity at the colorectal cancer invasive front. Cell 182, 1341–1359.e19 (2020).
141. Wieland, A. et al. Defining HPV-specific B cell responses in patients with head and
neck cancer. Nature 597, 274–278 (2021).
142. Gaglia, G. et al. Temporal and spatial topography of cell proliferation in cancer. Nat. Cell
Biol. 24, 316–326 (2022).
143. Kather, J. N. et al. Pan-cancer image-based detection of clinically actionable genetic
alterations. Nat. Cancer 1, 789–799 (2020).
144. Fu, Y. et al. Pan-cancer computational histopathology reveals mutations, tumor
composition and prognosis. Nat. Cancer 1, 800–810 (2020).
145. Coudray, N. et al. Classification and mutation prediction from non-small cell lung cancer
histopathology images using deep learning. Nat. Med. 24, 1559–1567 (2018).
146. Loeffler, C. M. L. et al. Artificial intelligence–based detection of FGFR3 mutational status
directly from routine histology in bladder cancer: a possible preselection for molecular
testing? Eur. Urol. Focus. 8, 472–479 (2022).
147. Kole, A. J. et al. Overall survival is improved when DCIS accompanies invasive breast
cancer. Sci. Rep. 9, 9934 (2019).
148. Ponz de Leon, M. et al. Evidence for the existence of different types of large bowel tumor:
suggestions from the clinical data of a population-based registry. J. Surg. Oncol. 44,
35–43 (1990).
149. Sawas, T. et al. Identification of prognostic phenotypes of esophageal adenocarcinoma
in 2 independent cohorts. Gastroenterology 155, 1720–1728.e4 (2018).
150. Ross-Innes, C. S. et al. Whole-genome sequencing provides new insights into the clonal
architecture of Barrett’s esophagus and esophageal adenocarcinoma. Nat. Genet. 47,
1038–1046 (2015).
151. Stachler, M. D. et al. Paired exome analysis of Barrett’s esophagus and adenocarcinoma.
Nat. Genet. 47, 1047–1055 (2015).
152. Thomlinson, R. H. & Gray, L. H. The histological structure of some human lung cancers
and the possible implications for radiotherapy. Br. J. Cancer 9, 539–549 (1955).
153. Hammerl, D. et al. Spatial immunophenotypes predict response to anti-PD1 treatment
and capture distinct paths of T cell evasion in triple negative breast cancer. Nat. Commun.
12, 5668 (2021).
154. Keren, L. et al. A structured tumor-immune microenvironment in triple negative
breast cancer revealed by multiplexed ion beam imaging. Cell 174, 1373–1387.e19
(2018).
155. Galon, J. et al. Towards the introduction of the ‘Immunoscore’ in the classification
of malignant tumours. J. Pathol. 232, 199–209 (2014).
156. Hegde, P. S. & Chen, D. S. Top 10 challenges in cancer immunotherapy. Immunity 52,
17–35 (2020).
157. Grünwald, B. T. et al. Spatially confined sub-tumor microenvironments in pancreatic
cancer. Cell 184, 5577–5592.e18 (2021).
311
Review article
158. Tavernari, D. et al. Nongenetic evolution drives lung adenocarcinoma spatial
heterogeneity and progression. Cancer Discov. 11, 1490–1507 (2021).
159. Rooney, M. S., Shukla, S. A., Wu, C. J., Getz, G. & Hacohen, N. Molecular and genetic
properties of tumors associated with local immune cytolytic activity. Cell 160, 48–61
(2015).
160. Thorsson, V. et al. The immune landscape of cancer. Immunity 51, 411–412 (2019).
161. McGranahan, N. et al. Clonal neoantigens elicit T cell immunoreactivity and sensitivity
to immune checkpoint blockade. Science 351, 1463–1469 (2016).
162. Shukla, S. A. et al. Comprehensive analysis of cancer-associated somatic mutations
in class I HLA genes. Nat. Biotechnol. 33, 1152–1158 (2015).
163. Goodman, A. M. et al. Prevalence of PDL1 amplification and preliminary response
to immune checkpoint blockade in solid tumors. JAMA Oncol. 4, 1237–1244 (2018).
164. Van den Eynden, J., Jiménez-Sánchez, A., Miller, M. L. & Larsson, E. Lack of detectable
neoantigen depletion signals in the untreated cancer genome. Nat. Genet. 51, 1741–1748
(2019).
165. McGranahan, N. et al. Allele-specific HLA loss and immune escape in lung cancer
evolution. Cell 171, 1259–1271.e11 (2017).
A seminal study that revealed ample evidence of subclonal immune evasion.
166. Balachandran, V. P. et al. Identification of unique neoantigen qualities in long-term
survivors of pancreatic cancer. Nature 551, 512–516 (2017).
167. Łuksza, M. et al. Neoantigen quality predicts immunoediting in survivors of pancreatic
cancer. Nature 606, 389–395 (2022).
168. Sade-Feldman, M. et al. Resistance to checkpoint blockade therapy through inactivation
of antigen presentation. Nat. Commun. 8, 1136 (2017).
169. Zaretsky, J. M. et al. Mutations associated with acquired resistance to PD-1 blockade
in melanoma. N. Engl. J. Med. 375, 819–829 (2016).
170. Dhainaut, M. et al. Spatial CRISPR genomics identifies regulators of the tumor
microenvironment. Cell 185, 1223–1239.e20 (2022).
A mouse experimental CRISPR model demonstrating direct effects of gene knockouts
and TMEs.
171. Vennin, C. et al. CAF hierarchy driven by pancreatic cancer cell p53-status creates
a pro-metastatic and chemoresistant environment via perlecan. Nat. Commun. 10, 3637
(2019).
172. Tape, C. J. et al. Oncogenic KRAS regulates tumor cell signaling via stromal reciprocation.
Cell 165, 1818 (2016).
173. Ali, H. R. et al. Imaging mass cytometry and multiplatform genomics define the
phenogenomic landscape of breast cancer. Nat. Cancer 1, 163–175 (2020).
174. Harris, A. L. Hypoxia — a key regulatory factor in tumour growth. Nat. Rev. Cancer 2,
38–47 (2002).
175. Graeber, T. G. et al. Hypoxia-mediated selection of cells with diminished apoptotic
potential in solid tumours. Nature 379, 88–91 (1996).
176. Baldominos, P. et al. Quiescent cancer cells resist T cell attack by forming an
immunosuppressive niche. Cell 185, 1694–1708.e19 (2022).
177. Miller, B. E., Miller, F. R. & Heppner, G. H. Therapeutic perturbation of the tumor
ecosystem in reconstructed heterogeneous mouse mammary tumors. Cancer Res. 49,
3747–3753 (1989).
178. Alonso-Curbelo, D. et al. A gene-environment-induced epigenetic program initiates
tumorigenesis. Nature 590, 642–648 (2021).
179. Cleary, A. S., Leonard, T. L., Gestl, S. A. & Gunther, E. J. Tumour cell heterogeneity
maintained by cooperating subclones in Wnt-driven mammary cancers. Nature 508,
113–117 (2014).
180. Zhou, H., Neelakantan, D. & Ford, H. L. Clonal cooperativity in heterogenous cancers.
Semin. Cell Dev. Biol. 64, 79–89 (2017).
181. Tammela, T. et al. A Wnt-producing niche drives proliferative potential and progression
in lung adenocarcinoma. Nature 545, 355–359 (2017).
182. Williams, J. B. et al. Tumor heterogeneity and clonal cooperation influence the immune
selection of IFN-γ-signaling mutant cancer cells. Nat. Commun. 11, 602 (2020).
183. Diamantopoulou, Z. et al. The metastatic spread of breast cancer accelerates during
sleep. Nature 607, 156–162 (2022).
184. Szczerba, B. M. et al. Neutrophils escort circulating tumour cells to enable cell cycle
progression. Nature 566, 553–557 (2019).
185. Adams, D. L. et al. Circulating giant macrophages as a potential biomarker of solid
tumors. Proc. Natl Acad. Sci. USA 111, 3514–3519 (2014).
186. Paget, S. The distribution of secondary growths in cancer of the breast. Lancet 133,
571–573 (1889).
187. Zeng, Q. et al. Synaptic proximity enables NMDAR signalling to promote brain
metastasis. Nature 573, 526–531 (2019).
188. Venkataramani, V. et al. Glioblastoma hijacks neuronal mechanisms for brain invasion.
Cell 185, 2899–2917.e31 (2022).
189. Quinn, J. J. et al. Single-cell lineages reveal the rates, routes, and drivers of metastasis
in cancer xenografts.Science 371, eabc1944 (2021).
190. Peinado, H. et al. Pre-metastatic niches: organ-specific homes for metastases. Nat. Rev.
Cancer 17, 302–317 (2017).
191. Shiozawa, Y. et al. Human prostate cancer metastases target the hematopoietic stem cell
niche to establish footholds in mouse bone marrow. J. Clin. Invest. 121, 1298–1312 (2011).
192. Kienast, Y. et al. Real-time imaging reveals the single steps of brain metastasis formation.
Nat. Med. 16, 116–122 (2010).
193. Ghajar, C. M. et al. The perivascular niche regulates breast tumour dormancy. Nat. Cell
Biol. 15, 807–817 (2013).
Nature Reviews Genetics | Volume 24 | May 2023 | 295–313
194. Fane, M. E. et al. Stromal changes in the aged lung induce an emergence from
melanoma dormancy. Nature 606, 396–405 (2022).
195. Reticker-Flynn, N. E. et al. Lymph node colonization induces tumor-immune tolerance
to promote distant metastasis. Cell 185, 1924–1942.e23 (2022).
196. van Maldegem, F. et al. Characterisation of tumour microenvironment remodelling
following oncogene inhibition in preclinical studies with imaging mass cytometry.
Nat. Commun. 12, 5906 (2021).
197. Moldoveanu, D. et al. Spatially mapping the immune landscape of melanoma using
imaging mass cytometry. Sci. Immunol. 7, eabi5072 (2022).
198. Erickson, A. et al. Spatially resolved clonal copy number alterations in benign and
malignant tissue. Nature 608, 360–367 (2022).
199. Azimi, F. et al. Tumor-infiltrating lymphocyte grade is an independent predictor
of sentinel lymph node status and survival in patients with cutaneous melanoma.
J. Clin. Oncol. 30, 2678–2683 (2012).
200. Lee, H. J. et al. Tertiary lymphoid structures: prognostic significance and relationship
with tumour-infiltrating lymphocytes in triple-negative breast cancer. J. Clin. Pathol. 69,
422–430 (2016).
201. Helmink, B. A. et al. B cells and tertiary lymphoid structures promote immunotherapy
response. Nature 577, 549–555 (2020).
202. Chen, B. et al. Prognostic value of the common tumour-infiltrating lymphocyte subtypes
for patients with non-small cell lung cancer: a meta-analysis. PLoS One 15, e0242173
(2020).
203. Salgado, R. et al. The evaluation of tumor-infiltrating lymphocytes (TILs) in breast cancer:
recommendations by an International TILs Working Group 2014. Ann. Oncol. 26, 259–271
(2015).
204. Lu, S. et al. Comparison of biomarker modalities for predicting response to PD-1/PD-L1
checkpoint blockade: a systematic review and meta-analysis. JAMA Oncol. 5, 1195–1204
(2019).
205. Galon, J. et al. Type, density, and location of immune cells within human colorectal
tumors predict clinical outcome. Science 313, 1960–1964 (2006).
206. Pagès, F. et al. International validation of the consensus Immunoscore for the
classification of colon cancer: a prognostic and accuracy study. Lancet 391, 2128–2139
(2018).
207. Färkkilä, A. et al. Immunogenomic profiling determines responses to combined PARP
and PD-1 inhibition in ovarian cancer. Nat. Commun. 11, 1459 (2020).
208. González-Silva, L., Quevedo, L. & Varela, I. Tumor functional heterogeneity unraveled
by scRNA-seq technologies. Trends Cancer Res. 6, 13–19 (2020).
209. Janiszewska, M. et al. Subclonal cooperation drives metastasis by modulating local and
systemic immune microenvironments. Nat. Cell Biol. 21, 879–888 (2019).
210. Gatenbee, C. D. et al. Immunosuppressive niche engineering at the onset of human
colorectal cancer. Nat. Commun. 13, 1798 (2022).
211. Weinstein, J. N. et al. The cancer genome atlas pan-cancer analysis project. Nat. Genet.
45, 1113–1120 (2013).
212. International Cancer Genome Consortium. International network of cancer genome
projects. Nature 464, 993–998 (2010).
213. ICGC/TCGA Pan-Cancer Analysis of Whole Genomes Consortium. Pan-cancer analysis
of whole genomes. Nature 578, 82–93 (2020).
214. Rozenblatt-Rosen, O. et al. The human tumor atlas network: charting tumor transitions
across space and time at single-cell resolution. Cell 181, 236–249 (2020).
215. Rao, A., Barkley, D., França, G. S. & Yanai, I. Exploring tissue architecture using spatial
transcriptomics. Nature 596, 211–220 (2021).
216. Emmert-Buck, M. R. et al. Laser capture microdissection. Science 274, 998–1001 (1996).
217. Ellis, P. et al. Reliable detection of somatic mutations in solid tissues by laser-capture
microdissection and low-input DNA sequencing. Nat. Protoc. 16, 841–871 (2021).
218. Nichterwitz, S. et al. Laser capture microscopy coupled with Smart-seq2 for precise
spatial transcriptomic profiling. Nat. Commun. 7, 12139 (2016).
219. Cui Zhou, D. et al. Spatially restricted drivers and transitional cell populations cooperate
with the microenvironment in untreated and chemo-resistant pancreatic cancer.
Nat. Genet. 54, 1390–1405 (2022).
220. Rodriques, S. G. et al. Slide-seq: a scalable technology for measuring genome-wide
expression at high spatial resolution. Science 363, 1463–1467 (2019).
221. Stickels, R. R. et al. Highly sensitive spatial transcriptomics at near-cellular resolution
with Slide-seqV2. Nat. Biotechnol. 39, 313–319 (2021).
222. Vickovic, S. et al. High-definition spatial transcriptomics for in situ tissue profiling.
Nat. Methods 16, 987–990 (2019).
223. Liu, Y. et al. High-spatial-resolution multi-omics sequencing via deterministic barcoding
in tissue. Cell 183, 1665–1681.e18 (2020).
224. Cho, C.-S. et al. Microscopic examination of spatial transcriptome using Seq-Scope. Cell
184, 3559–3572.e22 (2021).
225. Deng, Y. et al. Spatial-CUT&Tag: spatially resolved chromatin modification profiling at the
cellular level. Science 375, 681–686 (2022).
226. Deng, Y. et al. Spatial profiling of chromatin accessibility in mouse and human tissues.
Nature 609, 375–383 (2022).
227. Chen, A. et al. Spatiotemporal transcriptomic atlas of mouse organogenesis using DNA
nanoball-patterned arrays. Cell 185, 1777–1792.e21 (2022).
228. Chen, K. H., Boettiger, A. N., Moffitt, J. R., Wang, S. & Zhuang, X. RNA imaging. Spatially
resolved, highly multiplexed RNA profiling in single cells. Science 348, aaa6090 (2015).
229. Lyubimova, A. et al. Single-molecule mRNA detection and counting in mammalian
tissue. Nat. Protoc. 8, 1743–1758 (2013).
312
Review article
230. Chen, F. et al. Nanoscale imaging of RNA with expansion microscopy. Nat. Methods 13,
679–684 (2016).
231. Coskun, A. F. & Cai, L. Dense transcript profiling in single cells by image correlation
decoding. Nat. Methods 13, 657–660 (2016).
232. Codeluppi, S. et al. Spatial organization of the somatosensory cortex revealed
by osmFISH. Nat. Methods 15, 932–935 (2018).
233. Wang, X. et al. Three-dimensional intact-tissue sequencing of single-cell transcriptional
states. Science 361, eaat5691 (2018).
234. Lee, J. H. et al. Fluorescent in situ sequencing (FISSEQ) of RNA for gene expression
profiling in intact cells and tissues. Nat. Protoc. 10, 442–458 (2015).
235. Qian, X. et al. Probabilistic cell typing enables fine mapping of closely related cell types
in situ. Nat. Methods 17, 101–106 (2020).
236. Chen, X., Sun, Y.-C., Church, G. M., Lee, J. H. & Zador, A. M. Efficient in situ barcode
sequencing using padlock probe-based BaristaSeq. Nucleic Acids Res. 46, e22 (2018).
237. Alon, S. et al. Expansion sequencing: Spatially precise in situ transcriptomics in intact
biological systems. Science 371, eaax2656 (2021).
238. Gyllborg, D. et al. Hybridization-based in situ sequencing (HybISS) for spatially resolved
transcriptomics in human and mouse brain tissue. Nucleic Acids Res. 48, e112 (2020).
239. Su, J.-H., Zheng, P., Kinrot, S. S., Bintu, B. & Zhuang, X. Genome-scale imaging of the
3D organization and transcriptional activity of chromatin. Cell 182, 1641–1659.e26 (2020).
240. Lu, T., Ang, C. E. & Zhuang, X. Spatially resolved epigenomic profiling of single cells in
complex tissues. Cell 185, 4448-4464.e17 (2022).
241. Eng, C.-H. L. et al. Transcriptome-scale super-resolved imaging in tissues by RNA
seqFISH. Nature 568, 235–239 (2019).
242. Wang, F. et al. RNAscope: a novel in situ RNA analysis platform for formalin-fixed,
paraffin-embedded tissues. J. Mol. Diagn. 14, 22–29 (2012).
243. Ke, R. et al. In situ sequencing for RNA analysis in preserved tissue and cells. Nat. Methods
10, 857–860 (2013).
244. Baker, A.-M. et al. Robust RNA-based in situ mutation detection delineates colorectal
cancer subclonal evolution. Nat. Commun. 8, 1998 (2017).
245. Goltsev, Y. et al. Deep profiling of mouse splenic architecture with CODEX multiplexed
imaging. Cell 174, 968–981.e15 (2018).
246. Saka, S. K. et al. Immuno-SABER enables highly multiplexed and amplified protein
imaging in tissues. Nat. Biotechnol. 37, 1080–1090 (2019).
247. Lin, J.-R. et al. Highly multiplexed immunofluorescence imaging of human tissues and
tumors using t-CyCIF and conventional optical microscopes. eLife 7, e31657 (2018).
248. Lin, J.-R., Fallahi-Sichani, M. & Sorger, P. K. Highly multiplexed imaging of single cells using
a high-throughput cyclic immunofluorescence method. Nat. Commun. 6, 8390 (2015).
249. Gerdes, M. J. et al. Highly multiplexed single-cell analysis of formalin-fixed,
paraffin-embedded cancer tissue. Proc. Natl Acad. Sci. USA 110, 11982–11987 (2013).
250. Keren, L. et al. MIBI-TOF: a multiplexed imaging platform relates cellular phenotypes
and tissue structure. Sci. Adv. 5, eaax5851 (2019).
251. He, S. et al. High-plex multiomic analysis in FFPE at subcellular level by spatial molecular
imaging. Preprint at bioRxiv https://doi.org/10.1101/2021.11.03.467020 (2022).
252. Bergholtz, H. et al. Best practices for spatial profiling for breast cancer research with
the GeoMx® digital spatial profiler. Cancers 13, 4456 (2021).
253. Giesen, C. et al. Highly multiplexed imaging of tumor tissues with subcellular resolution
by mass cytometry. Nat. Methods 11, 417–422 (2014).
254. Angelo, M. et al. Multiplexed ion beam imaging of human breast tumors. Nat. Med. 20,
436–442 (2014).
255. Mund, A., Brunner, A.-D. & Mann, M. Unbiased spatial proteomics with single-cell
resolution in tissues. Mol. Cell 82, 2335–2349 (2022).
256. Mund, A. et al. Deep Visual Proteomics defines single-cell identity and heterogeneity.
Nat Biotechnol 40, 1231–1240 (2022).
257. Palla, G., Fischer, D. S., Regev, A. & Theis, F. J. Spatial components of molecular tissue
biology. Nat. Biotechnol. 40, 308–318 (2022).
258. Lin, J.-R. et al. Multiplexed 3D atlas of state transitions and immune interactions in
colorectal cancer. Preprint at bioRxiv https://doi.org/10.1101/2021.03.31.437984 (2021).
259. Gataric, M. et al. PoSTcode: Probabilistic image-based spatial transcriptomics decoder.
Preprint at bioRxiv https://doi.org/10.1101/2021.10.12.464086 (2021).
260. Dries, R. et al. Advances in spatial transcriptomic data analysis. Genome Res. 31,
1706–1718 (2021).
261. Perkel, J. M. Starfish enterprise: finding RNA patterns in single cells. Nature 572, 549–551
(2019).
262. Hickey, J. W. et al. Spatial mapping of protein composition and tissue organization:
a primer for multiplexed antibody-based imaging. Nat. Methods 19, 284–295 (2022).
263. Kleshchevnikov, V. et al. Cell2location maps fine-grained cell types in spatial
transcriptomics. Nat. Biotechnol. 40, 661–671 (2022).
Nature Reviews Genetics | Volume 24 | May 2023 | 295–313
264. Nieto, P. et al. A single-cell tumor immune atlas for precision oncology. Genome Res.
31, 1913–1926 (2021).
265. Park, J. et al. Cell segmentation-free inference of cell types from in situ transcriptomics
data. Nat. Commun. 12, 3545 (2021).
266. Petukhov, V. et al. Cell segmentation in imaging-based spatial transcriptomics.
Nat. Biotechnol. 40, 345–354 (2022).
267. Soldatov, R. et al. Spatiotemporal structure of cell fate decisions in murine neural crest.
Science 364, eaas9536 (2019).
268. Prabhakaran, S. Sparcle: assigning transcripts to cells in multiplexed images.
Bioinforma. Adv. 2, vbac048 (2022).
269. Schapiro, D. et al. MCMICRO: a scalable, modular image-processing pipeline
for multiplexed tissue imaging. Nat. Methods 19, 311–315 (2022).
270. Lohoff, T. et al. Integration of spatial and single-cell transcriptomic data elucidates
mouse organogenesis. Nat. Biotechnol. 40, 74–85 (2022).
271. Zhang, Y. et al. Reference-based cell type matching of spatial transcriptomics data.
Preprint at bioRxiv https://doi.org/10.1101/2022.03.28.486139 (2022).
272. Stuart, T. et al. Comprehensive integration of single-cell data. Cell 177, 1888–1902.e21
(2019).
273. Kiemen, A. et al. In situ characterization of the 3D microanatomy of the pancreas
and pancreatic cancer at single cell resolution. Preprint at bioRxiv https://doi.org/
10.1101/2020.12.08.416909 (2020).
274. Velten, B. et al. Identifying temporal and spatial patterns of variation from multimodal
data using MEFISTO. Nat. Methods 19, 179–186 (2022).
275. Svensson, V., Teichmann, S. A. & Stegle, O. SpatialDE: identification of spatially variable
genes. Nat. Methods 15, 343–346 (2018).
276. Lopez, R. et al. DestVI identifies continuums of cell types in spatial transcriptomics data.
Nat. Biotechnol. 40, 1360–1369 (2022).
277. McCaffrey, E. F. et al. The immunoregulatory landscape of human tuberculosis granulomas.
Nat. Immunol. 23, 318–329 (2022).
278. Dong, K. & Zhang, S. Deciphering spatial domains from spatially resolved transcriptomics
with an adaptive graph attention auto-encoder. Nat. Commun. 13, 1739 (2022).
279. Nagasawa, S. et al. Genomic profiling reveals heterogeneous populations of ductal
carcinoma in situ of the breast. Commun. Biol. 4, 438 (2021).
280. Elyanow, R., Zeira, R., Land, M. & Raphael, B. J. STARCH: copy number and clone
inference from spatial transcriptomics data. Phys. Biol. 18, 035001 (2021).
Acknowledgements
We would like to thank S. Dentro, I. Martincorena and T. Grünewald for critical feedback on
the manuscript. L.R.Y. is funded by a Wellcome Trust Clinical Research Career Development
Fellowship ref: 214584/Z/18/Z.
Author contributions
Z.S. and A.L. conducted the literature research. Z.S. drew the initial figures based on
discussions with all authors. M.G. conceived and supervised the study with L.R.Y. All authors
wrote the manuscript.
Competing interests
The authors declare no competing interests.
Additional information
Correspondence should be addressed to Moritz Gerstung.
Peer review information Nature Reviews Genetics thanks Alexander Anderson, Alexander
Swarbrick, who co-reviewed with Sonny Ramkomuth, and the other, anonymous, reviewer(s)
for their contribution to the peer review of this work.
Reprints and permissions information is available at www.nature.com/reprints.
Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in
published maps and institutional affiliations.
Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this
article under a publishing agreement with the author(s) or other rightsholder(s); author self-
archiving of the accepted manuscript version of this article is solely governed by the terms
of such publishing agreement and applicable law.
© Springer Nature Limited 2022
313