Genetics and Biomarkers in Barrett’s Esophagus and Esophageal Adenocarcinoma

4.2.2.1

Altered Cell Signaling

Normal growth and division of cells are carefully controlled. Cells require a variety of growth factors, with their associated intracellular signaling machinery, to allow cell division and prevent apoptosis while keeping abnormal division under check. It is this fine balance of pro- and antiproliferative signals that cancer cells must alter to allow their uncontrolled proliferation .

BE cells have dysregulated this balance, leading to increased proliferation , however they do not generally contain the genetic alterations known to cause pro-proliferative growth factor signaling that are seen in some other precursor lesions, such as in the pancreas where such precancerous lesions appear to be initiated by KRAS mutations in 90% of cases . However, there are direct effects of pulsatile pH and bile acids on cell cycle and the pro-inflammatory environment may also play a role . Inflammatory cytokines such as IL-8 are produced by the epithelial cells in response to reflux and these cytokines act via pathways such as STAT3 signaling which lead to proliferation, intended to replenish the epithelium . Other important inflammatory pathways are directly activated by reflux such as NF-κB . Reflux also causes the production of Reactive Oxygen Species (ROS) via various pathways including inhibition of mitochondrial electron transport in epithelial cells and production by infiltrating immune cells . These ROS species have several effects; they can induce DNA and protein damage, inducing the mutations required for the continued progression of BE , but can also affect signaling pathways which utilize endogenous ROS production to transmit signals. For instance in EGFR and PDGFR signaling, known to promote cell proliferation and carcinogenesis, the negative regulator PTEN can be specifically, reversibly inhibited by ROS causing increased proliferation and inhibition of apoptosis . As well as causing an inflammatory response, reflux is thought to cause changes in cell and tissue morphology that are associated with BE. There is accumulating evidence that BE tissue has acquired a greater resistance to the reflux damage . Hence it is presumed that death of cells due to reflux damage combined with inflammation associated proliferation provides an environment in which a subset of cells in the vicinity that are better able to survive reflux, either due to genomic alterations or a different differentiation program, come to predominate the tissue.

As well as the damage induced by inflammation and reflux, somatic alterations in BE appear to drive clonal expansion . Genetic changes in BE are very common but as usual in the development of cancer, most of these gene mutations are passengers rather than being causal in pathogenesis. Genes in which mutations appear to be selected above this background rate in BE include CDKN2A (p16) and TP53 (p53).

P16 is a small protein which binds to and inhibits cyclin dependent kinase 4 (CDK4) and CDK6, thereby inhibiting the phosphorylation of Rb, and preventing cell cycle progression and cell division . It is activated by stimuli such as DNA damage and ROS , both caused by reflux, and hence it is likely that loss of this cell cycle inhibitor allows a greater rate of cell division in this environment. CDKN2A is commonly lost in BE either by mutation and loss of heterozygosity (LOH) or epigenetic silencing in approximately 15% and 60% of patients respectively and these genetic changes are associated with expansions by spatial mapping of cell clones on the surface of BE . P53 is a transcription factor which acts to inhibit cell proliferation and activate apoptosis via regulation of a variety of other genes. It is also activated in response to DNA damage and ROS. P53 function is lost in many different cancers at a high rate and hence may be a particularly vital node in these tumor-suppressive signaling pathways . P53 mutation is relatively uncommon in Non-Dysplastic BE (NDBE) but prevalent in HGD occurring in approximately 86% of patients . This occurs mostly via point mutations, is commonly accompanied by LOH, and is also associated with clonal expansion .

Evidence for specific gene mutations that demarcate the boundary between HGD and EAC, and hence may be important in this transition, is difficult to find. This was demonstrated in a recent study by Weaver and colleagues where Targeted Sequencing was used to compare single nucleotide variant (SNV) mutations in cases of NDBE, HGD, and EAC with a stable phenotype and demonstrated dramatic heterogeneity across all three states. Weaver et al. identified only one gene mutation, SMAD4 , which consistently associated with EAC rather than BE, and only at a low rate (13%). SMAD4 is a central component for signaling via ligands of the transforming growth factor beta (TGFβ) superfamily . SMAD4 is commonly lost in many other cancers, for instance in pancreatic carcinoma where it is lost in 31% of cases and carcinogenesis has been associated with a switch in TGFβ signaling from antiproliferative SMAD4 dependent TGFβ signaling to invasion and migration of cells via SMAD4 independent signaling . The rate of point mutation in p53 is already high in HGD and did not increase with progression to EAC, but it was maintained . TP53 LOH and altered protein expression are associated with an increased risk of progression (see Table 4.3 ) and have pro-invasive and pro-migratory affects in vitro . It is therefore thought that loss of TP53 function is required for progression to EAC in the majority of patients. This is further evidenced by the mutation’s effect on genomic stability , the importance of which in progression to EAC shall be discussed subsequently.

This genetic heterogeneity demonstrated by Weaver et al. has several possible explanations: it is possible that only a very few genetic changes are required for the transition to EAC, that other genomic changes, not detected in this study, such as large-scale chromosomal rearrangements-, copy number alterations or SNVs in genes not targeted in this study drive the process; or that, as discussed, whole gene networks are being targeted by such mutations rather than many individual genes. Hence pathway analysis of whole gene networks, using a wider variety of genomic alterations, in larger cohorts of BE and EAC could give a better indication of how this transition occurs, but may be limited by our incomplete understanding of these extremely complex networks. The genes in which genetic alterations occur have helped us little in understanding the molecular changes which underlie the development of EAC as yet, however more clues are perhaps to be found in the types of genetic alteration that occur.

4.2.2.2

Mechanisms of Genetic Alteration

Loss of genomic integrity is a vital constituent of the cancer phenotype . Recent advances in next-generation sequencing technology have allowed the identification of a greater variety of genetic alterations and the increasing affordability has allowed larger patient cohorts to be investigated.

Millions of SNVs have been analyzed across thousands of patients in Many different cancers, and this has allowed statistical analyses to identify patterns termed mutational signatures . These consist of biases in base conversions that occur in particular immediate sequence contexts, for instance it is common to find C-T conversions 5’ to a G nucleotide in many cancers due to deamination of Methyl-CpG dinucleotides. These signatures vary both between different cancers and between patients. In EAC an unusual signature of AA-CA conversions, with a preference for 5’ G nucleotides, has been identified , alongside other more common signatures. Some signatures appear to be associated with either specific types of damage, smoking for instance, or with mutations in particular DNA repair genes, such as BRCA2 . Overall there is a high number of SNVs in EAC, comparable only with cancers driven by specific mutagens such as melanoma or lung cancer . This caused some to suggest that this mutational signature could be due to reflux-associated oxidative stress in BE .

Deletions, insertions, inversions and translocations effecting large genomic regions are common in cancer, in particular EAC , and are collectively known as structural variants (SVs). With the resolution of new sequencing technologies it is becoming apparent that these changes account for a significant number of tumor suppressor inactivation events in EAC . Deletions or amplifications occur in large sections of chromosomes, whole chromosomes and even the whole karyotype, amplifying and deleting oncogenes and tumor suppressors. This is far easier to detect and has been known for many years . A recent study identified such large-scale genomic variation as an important marker for the transition to EAC . They identified changes in copy number across the genome in a longitudinal study of 248 BE patients. In patients who did not progress to EAC copy number alterations did not significantly alter over time–however, in the 79 patients who did progress, the mean number of copy number alterations increased rapidly beginning approximately 24 months before EAC was diagnosed. Importantly, these patients with high structural variability were still histologically diagnosed as BE during this 24-month period and hence this gain of large-scale genomic instability appears to be an important precursor step in the transition to EAC in many patients. As the degree of dysplasia in this study was not commented on we cannot be sure how this precursor step relates to the pathological state. Copy number changes lead to amplification of known oncogenes such as MYC , ERBB2 , EGFR , and KRAS , however the only copy number change statistically associated with progressors rather than nonprogressors in this study was still deletion at the SMAD4 locus. Whether this genomic instability, perhaps induced by TP53 mutation which also occurs late in the progression of BE at the stage of HGD , is causally involved in progression or simply a consequence of changes which themselves lead to EAC development is unknown.

The spatial and temporal genomic distribution of mutations has long been presumed to be random, that is mutation events occur fairly evenly across the genome, even if they are then concentrated via selection, and in an independent manner over many cell generations. However local hypermutation of both SNVs and SVs has been identified in EAC, phenomena termed kataegis, and chromothripsis respectively and there is evidence that chromothripsis may be due to single catastrophic events. Chromothripsis, Greek for “Chromosome Shattering,” is thought to occur via a dramatic break event with currently unknown stimuli where the locus is broken into multiple pieces, and then stitched back together by the DNA repair machinery . Variations in copy number state within these loci tend to be limited which has led many to the conclusion that the events occur at a single point in time . Such events are not frequent in comparison to other types of mutation. Single chromothriptic-like events are only detected in 36% of tumors and 82% of tumors have fewer than 10 kataegic foci . The importance of these events, relative to gradual accumulation of SVs and SNVs across the genome, is unknown.

Modification of gene function in cancer is not only achieved by alteration of base sequence, as has thus far been discussed, but also by other modifications both in DNA itself and the packaging proteins, histones. Such modifications are used in normal somatic cells to regulate gene expression, and are particularly important during development to allow differentiation . These modifications consist of methylation directly onto DNA and a variety of chemical modifications that occur on specific residues of histone proteins . These changes alter how DNA in these regions are packaged and so affects the availability of genes contained within these regions to RNA polymerases. Recent technological developments have allowed the DNA methylation profile of whole genomes to be assessed. Such assessment of histone modifications on a whole genome scale is much more difficult but is likely to be equally important in cancer development . Hypermethylation of CDKN2A is common, as discussed, and clonally selected, and so possibly important in the pathogenesis of BE but such genome-wide approaches have identified multiple other genes which show similar patterns of hypermethylation such as APC, ESR1, REPRIMO, and many others . Global hypomethylation is also a feature of BE and EAC, as in many cancers, and results in upregulation of genes perhaps important in BE pathogenesis . However many of these studies define aberrant methylation relative to squamous cells in the esophagus. These differences may therefore not all be important in the pathogenesis of BE but simply be fundamental differences in the epigenetic differentiation program between squamous and columnar tissues types . Molecular analyses have provided insight into BE progression ( Fig. 4.1 ). However, further work is required if we wish to predict and prevent progression in the clinic.