What the cost of cells with detrimental mutations.

What is tumor heterogeneity, what are
the causes of tumor heterogeneity, and what are the implications of tumor



We Will Write a Custom Essay Specifically
For You For Only $13.90/page!

order now

“Tumor heterogeneity” refers
to the multiple populations of cells within a single tumor or in different
tumors that are unique in various characteristics such as signaling networks, growth
rates, cell surface properties, immunogenicities, and sensitivity to various
therapeutic agents (1). It is important to note that the concept of heterogeneity
applies to malignant cells within a primary tumor or single metastasis (intratumoral
heterogeneity) and between the primary tumor and metastases or among different
metastases (intertumoral heterogeneity) (1).


Genetic Causes of
Intratumoral Heterogeneity                                                                                         

Two models of genetically driven
intratumoral heterogeneity exist: the clonal evolution model and the cancer
stem cell (CSC) model (2). In the clonal evolution model, all
cancer cells in a tumor are equally tumorigenic. These cancer cells are genetically
unstable, leading to a wide variety of genetic variation during tumor
progression. Instability results from inherent errors in deoxyribonucleic acid
(DNA) replication that are exacerbated by uninhibited growth, from endogenous
and exogenous mutagens, and from deficiencies in DNA repair. Endogenous attacks
on DNA integrity include oxidative damages, depurinations, depyrimidinations,
single-strand breaks, double-strand breaks, and cytosine deaminations (3). Additionally, the exogenous attacks
include ultraviolet radiation, tobacco combustion products, and exposure to a
variety of rarer mutagens such as aflatoxin (3). Failures in the DNA repair machinery to
address these attacks can result in mutations. While cancer cells occasionally
harbor mutations in DNA repair machinery genes, they more frequently suppress the
expression of these genes to promote genetic instability and facilitate the
introduction of genetic variation. For example, Halford
et al. demonstrated that only 6/113 sequenced colorectal cancers had mutations
in the DNA repair gene MGMT but
methylation of the promoter region of MGMT
resulted in suppression of MGMT
expression in over half of the samples (4). However, not all mutations introduced
via genetic instability are beneficial to cancer cells. Cancer cells are
subjected to host selective pressures, including limitations in oxygen and
nutrients and encounters with tumoricidal immune cells. As tumors progress, the
selective pressures act on the variability in malignant cells, resulting in the
expansion of clones of cells with beneficial mutations at the cost of cells
with detrimental mutations. This Darwinian process, known as “clonal
expansion,” selects for the fittest cells to dominate the tumor, promoting
tumor growth and malignancy, which is a critical implication of tumor heterogeneity
as it strongly influences prognosis. Both linear and branching patterns of clonal
evolution exist. In the linear model, sequentially ordered mutations
accumulate, resulting in a series of clonal expansions in which a single clone
dominates the tumor until being overtaken by another clone. However, the linear
model frequently oversimplifies the process of tumor progression. In contrast,
the branching model involves a splitting mechanism that results in the
expansion of multiple subclonal populations following selection events. The end
result of the branching model is a heterogeneous tumor characterized by several
dominant subclones and numerous rare subclones. High-grade B cell lymphoblastic
leukemia is a cancer type believed to adhere to this model (2).


In contrast to the
clonal evolution model, the classical CSC model proposes that only a fraction
of the cancer cells in a tumor are actually tumorigenic (2). These self-renewing, undifferentiated cancer stem cells
(CSCs) divide infrequently but when they do, they divide asymmetrically to
produce a more differentiated transit-amplifying cell and an identical CSC (5). The transit-amplifying cells of a tumor divide rapidly
and continue to differentiate, generating the bulk of the tumor. Heterogeneity
results from this mix of tumorigenic and non-tumorigenic cells. Medulloblastomas,
malignant germ cell cancers, and glioblastomas are prominent examples of cancer
types that have been shown to adhere to the CSC model (2). However, the clonal differentiation model and classical CSC
model are not mutually exclusive. Indeed, it is highly conceivable that a tumor
could consist of several dominant clones organized in a hierarchical manner. In
this system, CSCs would accumulate mutations over time, introducing genetic
diversity into the system. Only beneficial mutations would lead to an expansion
of more differentiated progeny, resulting in a tumor consisting of several
dominant populations of cells derived from genetically heterogeneous CSCs (6).


The dynamic CSC model
varies from the classical CSC model by proposing that CSCs and non-CSCs interconvert, with the bidirectional
balance driven either by stochastic mutations or signals from the tumor
microenvironment (TME) (7).  Similar to the classical model of CSCs, the
dynamic model is highly compatible with the clonal selection model. Numerous
lines of evidence from melanoma and breast cancer research support the dynamic
model (7).
For example, Chaffer et al. found that ZEB1 mediated the conversion of
basal-type non-CSCs to CSCs in a model of breast cancer (8).


Non-Genetic Causes of
Intratumoral Heterogeneity

Regardless of the
genetic cause of intratumoral heterogeneity, variability in the TME also
introduces heterogeneity into tumors. Different
regions of a tumor are characterized by unique vasculature, stromal
infiltrates, and extracellular matrix. Therefore, malignant cells in a tumor
experience a range of cues from the TME, which in turn translate into a range
of phenotypes (1). Additionally,
alterations to the epigenome can contribute to variation in the phenotypes of
cells in different regions of a tumor. For example, the histone demethylase
KDM5A is at least partly responsible for a chromatin remodeling state
associated with significant resistance to EGFR inhibitors in a subpopulation of
non-squamous cell lung carcinomas. Treatment with EGFR inhibitors enriches this
subpopulation and sensitizes them to treatment with IGF-1 receptor inhibitors (9).



focused on analyzing the heterogeneity of metastases have frequently sought to
determine the relationship between primary tumors and metastases. Numerous
studies have demonstrated close relationships between metastases and their
primary tumors, suggesting that they are the final step of tumor progression.
For example, Bissig et al. identified a high probability of a common clonal
progenitor in 11 of 19 matched pairs of primary and metastatic renal cell
carcinomas (10). Additionally,
numerous gene expression studies have demonstrated significant similarity
between matched primary tumors and distant metastases (1,11-13). However, high degrees
of divergence have also been reported in numerous cases. For example, radically
different patterns of allelic losses, indicative of a high degree of genetic
divergence, have been reported between primary tumors and lymph node metastases
in prostate cancers (14).


evolution must also be considered when assessing intertumoral heterogeneity.
Metastases can differ significantly from each other after being shaped by their
microenvironments. Differences in the TMEs impose different selective pressures
on tumor cells, which can result in distinct subclones dominating in different
metastases within the same patient. For example, Stoecklein et al. demonstrated
that lymph node and bone marrow metastases from esophageal cancer patients substantially
diverged from primary tumors and each other, suggesting early metastasis and parallel
evolution shaped by TME-specific selection forces (15). Brastianos et al. determined that brain metastases
(BMs) mainly from breast, lung, and renal cell carcinoma patients diverged
significantly from patient-matched non-BMs but that multiple BMs from the same
patient were highly similar (16).


Methods to Analyze
Tumor Heterogeneity

Assays utilized to
study tumor heterogeneity fall into two categories: focused and global. Focused
assays are simple to interpret because they typically focus on functionally
important genes. In doing so, however, these assays neglect important
information relevant to tumor heterogeneity. Genome-wide approaches produce
unbiased, comprehensive results but can be difficult to interpret because of
the volume of data generated (1).


assays to identify tumor heterogeneity include allelic imbalance
identification, fluorescent in situ hybridization
(FISH), and sequencing of specific genes. Detection of allelic imbalances
involves isolating DNA from macrodissected tumor tissue and germline controls.
Polymerase chain reaction (PCR) is subsequently performed on the microsatellite
regions of the DNA. By comparing the results to germline controls, it is
possible to detect alleles that have been lost or gained during the evolution of
a tumor. Due to the poor correlation between deletions and amplifications in
the genome and changes in microsatellite markers, this technique typically
requires a second confirmatory method such as FISH analysis. In this technique,
researchers target regions of interest with fluorescently labeled probes, which
then hybridize to sequences of interest. While very laborious, it is a highly
validated technique for the detection of genomic regions amplified during tumor
progression. Finally, PCR is used to sequence specific genes implicated in the
progression of certain cancer types (1).


approaches to study tumor heterogeneity include karyotype analysis, comparative
genomic hybridization (CGH), single nucleotide polymorphism (SNP) arrays, and whole
genomic/exomic sequencing. Due to low resolution and inefficiency, karyotype
analysis is only used to detect large chromosomal aberrations. Many recent
publications have relied on CGH to detect chromosomal aberrations in an
unbiased manner. In this technique, specialized computational methods identify chromosomal
losses and gains following hybridization of a global array of fluorescent
probes to metaphase chromosomes, providing an unbiased assessment of the entire
genome (1). By hybridizing probes to an array
instead of metaphase chromosomes, SNP arrays are able to detect single-nucleotide
imbalances (1). Sequencing tumors provides a plethora
of information that can be used to study tumor heterogeneity. Sequencing
strategies include pooled sequencing, single-cell sequencing, and section
sequencing. Pooled sequencing analyzes DNA extracted from an entire tumor.
While the easiest to perform technically, this approach significantly limits
the inability to detect mutations in rare
subclones and to determine which subclones contain each mutation. However, it
can still provide useful information on intertumoral heterogeneity, as demonstrated
by Brastianos et al.’s use of this technique to analyze evolutionary
relationships between patient-matched primary tumors and BMs (16). In contrast, single-cell sequencing provides researchers
with the means to fully characterize individual cells, eliminating uncertainty
about which subclones have acquired a mutation of interest. Computational
algorithms such as SCITE, OncoNEM, and
SiFit allow for the construction of complex phylogenetic trees that map the
evolution of tumors and highlight which mutations confer advantages to cancer
cells within a tumor (17). However, it is still difficult with existing technology to adequately
power single-cell sequencing experiments. Section sequencing involves dividing
a tumor into sections and sequencing each section individually, followed by
algorithms such as PyClone, Clomial, and TargetClone. Compared to single-cell
sequencing, researchers can more readily identify which clones dominate
throughout space in the tumor and can more easily power their studies (18).  However, this
technique pools all clones together in a given area and is subject to the same limitations
as pooled sequencing for the area sampled.


Implications of Tumor Heterogeneity

90% of cancer-related deaths are the result of metastases resistant to therapy (19). Tumor heterogeneity significantly complicates
the treatment of these lesions. Cancer therapy constitutes a unique selection
pressure in the evolution of a tumor. In the clonal selection model, resistance
reflects selection of pre-existing subclones with resistance conferring
mutations. For example, Keats et al. utilized a collection of FISH arrays on 7 samples
serially collected from the same multiple myeloma patient at diagnosis,
remission following treatment with lenalidomide/dexamethasone,
and relapse on four subsequent treatment strategies. Their findings demonstrated
that two subclones present in the pretreatment sample alternated in extent of
dominance following each therapy until the disease became terminal, at which
point a subclone exceedingly rare in the pretreatment sample expanded to become
the dominant subclone (20). In contrast, the CSC model proposes
that CSCs mediate relapses to treatments. CSCs are remarkably resistant to a
variety of therapeutic modalities. Among several different mechanisms, CSCs mediate
resistance to chemotherapies by activating ATP-binding
cassette (ABC) transporters (21). A regimen of
chemotherapy might decimate the bulk of the tumor. However, relapse will occur
if CSCs expel the drugs and survive long enough to differentiate into highly
proliferative transit-amplifying cells. Regardless of the model, tumor heterogeneity
significantly complicates the care of patients, as indicated by the positive
correlation between tumor heterogeneity and worse overall survival in head and
neck cancers, acute myeloid leukemia, ovarian cancers, and lung cancers (22). To
further complicate matters, primary tumors differ considerably from metastases in
some cancer types. Different lesions might not respond the same to a given
treatment. For example, Higashiyama et al. demonstrated that lung cancer cell
lines derived from distant metastases were significantly less sensitive to
several chemotherapeutic agents than cell lines derived from primary tumors (23). Tumor heterogeneity requires a unique
approach to therapy. In addition to considering the sensitivity of each lesion
to a given regimen, it is necessary to provide consistent follow-up to manage
the emergence of resistant subclones.


addition to therapy, tumor heterogeneity has implications in biomarker
selection (24). A particular biomarker may not be
expressed by all cells in a single primary tumor or metastatic lesion, and the
markers may differ between metastatic lesions or between the primary tumor and
the metastatic lesions. However, tests that analyze the extent of heterogeneity
in a tumor provide useful information because of the inverse correlation
between heterogeneity and overall survival (22).


populations of malignant cells interact with each other in ways similar to
those described in ecology research: competition (one group benefits at the
cost of another group), mutualism (two or more groups cooperate in ways that
are beneficial for both groups), and commensalism (one group benefits without hurting
another group) (1). These interactions have significant implications for
tumor progression. Metabolic symbiosis is an example how mutualism promotes
tumor progression. In this system, hypoxic malignant cells utilize glycolysis
for their bioenergetic needs and secrete lactate into the TME via MCT4. Aerobic
cancer cells uptake the lactate via MCT1 and oxidize it to meet their
bioenergetic needs. This system allows cells facing different selective
pressures to behave in a mutually beneficial manner. Interestingly, Sonveaux et al. demonstrated that inhibition of MCT1 triggered
the aerobic cells to consume glucose in lieu of lactate, and the hypoxic cancer
cells died from glucose deprivation.


Tumor heterogeneity
has proven challenging for oncologists, and we must focus our attention on
understanding the nuances of this biological phenomenon in order to design more
effective treatments for patients.