Unlike sporadic colorectal cancer, colitis-associated cancer does not usually follow a linear pathway of gene loss or disruption but rather appears to be driven largely by inflammation-associated damage. Combined with the unclear etiology of this disease, researchers are pursuing multiple independent avenues to identify sensitive and specific biomarkers for early UC cancer. While there are currently many biomarkers in development to discriminate subtypes of patients with inflammatory bowel disease (IBD-ulcerative colitis vs Crohn’s disease), there are few biomarkers for differentiating UC non-Progressors (patients who do not develop dysplasia) from UC Progressors (patients who develop dysplasia or cancer). We will discuss some of the major efforts here. Due to limited space of the review, there will be a primary focus on markers with the most promising clinical potential. Table 1 summarizes the biomarker candidates from recent studies according to the types of biospecimens that have been used including colonic tissue, blood, urine and stool. Since many of these studies were performed on limited numbers of patient samples, they will need to be independently validated in multi-center larger cohort studies.
Biomarker development for UC-associated cancer has, for the most part, focused on the use of colonic biopsies obtained during colonoscopy or colectomy due to (1) the known pancolonic abnormalities that precede CAC; (2) colonic biopsies are a readily available source of material; and (3) retrospective material can be acquired. Immunohistochemical staining of formalin-fixed paraffin embedded (FFPE) material has provided information about disease etiology and clues to possible biomarker candidates. Isolation of genetic material (i.e., DNA, RNA, miRNA) as well as proteins for proteomic analysis from FFPE sections or frozen material has taken this to the next level. The hope is to identify alterations associated with UC cancerous or dysplastic regions that are also present in non-neoplastic and remote regions, ideally rectum. Rectal biopsies could be obtained with any laxative preparation, without sedation, and using a quick and minimally invasive procedure (proctoscopy) that could be performed in any clinic.
DNA: DNA-based assays using genetic material isolated from purified colonic epithelium has shown promise as a biomarker source. However, since UC doesn’t follow the hallmark loss of gene pathways like sporadic colorectal cancer, several studies have focused on more general genome-wide (rather than locus specific) instability. Pan-colonic chromosomal instability has been shown to be detectable relatively early in UC disease progression by BAC array[18,31], fluorescence in-situ hybridization (FISH)[19,20] and arbitrarily primed PCR (AP-PCR)[15,16], yet none of these assays are amenable to high throughput analysis that would be required for a clinical UC neoplastic biomarker. Clonal expansions of poly glutamine repeats (PolyG) are detectable in DNA isolated from colonic epithelium relative to adjacent stromal or muscle DNA controls. The poly G assay was validated in two independent cohorts which confirmed discrimination between UC Progressors and Non-progressors, and demonstrated its utility for UC patients with early disease onset (i.e., diagnosis before 50 years of age)[30,32]. A more recent study examined chromosomal instability in UC neoplastic progression using copy number variation microarrays, and found increased numbers of copy number variations with progression of UC to UC-associated colorectal cancer.
UC can also be thought of as an aging disease of the colon; qFISH and qPCR of colonic epithelium from UC patients showed accelerated telomere shortening compared to leukocyte or adjacent stromal controls[34,35]. Shortened telomeres result in chromosomal instability due to higher frequencies of anaphase bridges and subsequent chromosomal breakage, losses, and gains. These techniques lack either the very high specificity demanded by a CAC biomarker (qPCR) or capability for high throughput (qFISH) to create a clinically useable test.
With the development of whole-genome sequencing, comprehensive investigation of the genomic landscape has identified an increasing array of mutational signatures associated with specific diseases. In CAC, the chronic oxidative injury can cause mounting epithelial cell DNA damage which over time, overwhelms the G1 cell cycle checkpoint and results in p53 mutation. A recent study investigated somatic mutation patterns from inflammatory bowel disease (IBD, including ulcerative colitis and Crohn’s disease) associated with colorectal tumor using whole-exome sequencing. Not surprisingly, TP53 was the most commonly mutated gene, with mutational prevalence similar to that of sporadic colorectal tumors (63% of cases). However, APC and KRAS were mutated at significantly lower rates in tumors from patients with IBD than in sporadic colorectal tumors (13% and 20% of cases, respectively). Several IBD-specific gene mutations, including SOX9, EP300, NRG1, and IL16 were identified, confirming the notion that IBD association colorectal cancer (CRC) has its unique genetic composition compared to sporadic CRC.
RNA: Gene expression analysis using RT-PCR of materials isolated from colonic biopsy has resulted in identification of genes which are involved in or associated with UC neoplasia. Microarrays have been used to investigate gene expressional changes in UC neoplastic progression by comparing the mucosa of non-dysplastic, dysplastic and cancerous colonic tissues. The study identified genes associated with UC dysplasia and UC cancer. Five genes were further verified to be common to UC dysplasia and adenocarcinoma relative to non-dysplastic UC (CCND1, SERPINB6, PAP, IL8, and NOS2A).
Because the dysplasia or cancer in UC patients can appear as flat mucosa and/or is multi-foci, there is an interest in identifying neoplastic biomarkers that are present in both dysplastic and non-dysplastic tissues. Such biomarkers could then be tested in random biopsies, including rectal which could be obtained with a minimally invasive procedure. Numerous studies have identified gene expression changes in non-dysplastic tissues from UC Progressors. In one study, a gene signature which included ACSL1, BIRC3, CLC, CREM, ELTD1, FGG, S100A9, THBD, and TPD52L1 were progressively increased with neoplastic progression. These findings support the concept of a field defect phenomena in CAC. Two markers (S100A9 and REG1α) were further validated by IHC showing increased staining in the dysplastic and non-dysplastic tissues from UC Progressors compared to UC Non-progressors and normal controls. In a separate study, by surveying the expressions of 189 carcinogenesis related genes in non-dysplastic rectal mucosa from UC Progressors and Non-progressors, researchers identified a panel of 20 genes (including cancer genes such as CYP27B1, RUNX3, SAMSN1, EDIL3, NOL3, CXCL9, ITGB2, and LYN) in rectal tissues as Progressor associated genes. Using non-dysplastic rectal tissues, the 20-gene panel was able to predict UC-cancer patients from UC-non cancer patients with 83% accuracy and a negative predictive value of 100%.
DNA methylation: Promotor methylation plays an important role in tumorigenesis though transcriptional silencing of critical genes. The early study of DNA methylation in UC dysplasia could be traced back to a study investigating methylation status of five age or cancer related genes in UC dysplasia. Hypermethylation of ER, MYOD, p16, and CSPG2 were detectable in the high-grade dysplasia and cancer tissues from UC Progressors. Moreover, hypermethylation of ER, MYOD and p16 could be detected in the non-dysplastic tissues from Progressors compared to Non-progressors. In another study using methylation-specific PCR, hypermethylation of the tumor growth genes, FOXE1 and SYNE1, was detected in approximately 60%-80% of cancer samples (whereas it was undetectable in controls), suggesting an increase in methylation with disease progression. Methylation of the eyes absent homolog 4 (EYA4) gene was present both in neoplastic and remote non-neoplastic tissue of UC patients with cancer but absent in UC control patients without neoplasia[27,42]. Another study found altered methylation status of RUNX3, MINT1, and COX-2 in both the non-neoplastic regions and neoplastic regions of UC -CRC colons as compared with that in the UC controls. Either RUNX3 or MINT1 showed interaction with COX-2 with an additive effect. Further study is needed to evaluate whether this three-gene panel can predict the likelihood of patients to progress to CRC and/or to identify patients with dysplasia. While CpG island hypermethylation of p14 (ARF) but not p16 (INK4) was detectable in 100% of UC dysplasia tested, only 20% (2/10) of patients with dysplasia showed hypermethylation in DNA extracted from non-dysplastic rectum. In a recent study, the reduced expression of a tight junction-associated protein, BVES in UC neoplasia was shown to be the result of promotor hypermethylation of this gene. The BVES promoter hypermethylation could be detected in dysplastic colonic tissues as well as distant non-malignant-appearing mucosa from UC Progressors in comparison to UC Non-progressor and non-UC controls. The study further suggests that BVES interacts with PR61α to promote inflammatory tumorigenesis through c-Myc destruction. Based on the results, BVES promoter hypermethylation status could be a potential biomarker to identify patients with UC at risk of cancer.
MicroRNA: MicroRNAs are small non-coding 20-25 nucleotide single-stranded RNA molecules which usually bind to the 3’ untranslated region of target mRNA transcripts, effectively silencing them by inhibition of translation or through degradation. MicroRNAs can function as oncogenes to enhance cellular proliferation and survival or as tumor suppressors. Downregulation of miR-143 and miR-145, as well as concomitant upregulation of their predicted targets, IRS-1, K-Ras, API5, and MEK-2 was found in colon biopsies of UC patients, suggesting that some microRNAs could contribute directly to transformation of UC colonic epithelium. In colitis, the inflammatory microenvironment can modulate microRNA expression and further influence target gene expressions. Currently, there is much focus on the investigation of microRNAs that affect immune response in order to maintain intestinal homeostasis. MicroRNA analysis of colonic biopsies revealed decreased expression of 3 microRNAs (including miR-192, miR-375, and miR-422b) in active UC whereas upregulation of 8 microRNAs (miR-16, miR-21, miR-23a, miR-24, miR-29a, miR-126, miR-195, and Let-7f) was noted in active UC compared to normal controls. Upregulation of miR-21 and miR-155 in inflamed colon from UC patients compared to controls was confirmed in an independent study. Mouse studies suggest that miR-155 is involved in the proinflammatory cellular response due to decreased numbers of CD4+T, Th1/Th17, CD11b+, and CD11c+ cells in miR-155-/- mice. Dysregulation of some microRNAs could be related to inflammation and thus could be used to discriminate colitis associated cancers from sporadic colorectal cancers. For example, miR-26b was shown upregulated in CAC, and down-regulated in sporadic colon cancer.
Protein: Protein-based research has revealed some interesting CAC biomarker candidates. Protein participants in major known pathways involved in UC tumorigenesis have been further scrutinized to evaluate their potential as biomarkers. While p53 mutations have been shown to occur as an early event in UC neoplasia, the value of the mutated p53 protein as a clinical marker by IHC is controversial. The results of IHC studies are variable and limited due to the use of different antibodies and small sample sizes per study. Protein expression of p53 and chromogranin A provides moderate sensitivity (66.7%) and specificity (80%) for HGD detection. Other protein expression data shows that co-detection of p53 and α-methylacyl coenzyme A racemase (AMACR), using IHC, could discriminate UC patients who had dysplasia or cancer from those who did not; and further, the co-expression of p53 and AMACR was detectable in biopsies taken as early as 10-14 months prior to dysplasia. In a different study, expression of the apoptosis gene, Bcl-xl, is absent in non-dysplastic UC epithelium, but highly positive in dysplasia or cancer samples by IHC. Nuclear IHC staining of the programmed cell death 4 (PDCD4) tumor suppressor showed a reduction in colonic UC dysplasia and cancer tissue samples, and could be useful as a biomarker in the histological assessment of IBD-associated dysplastic lesions. Mitochondrial alterations as defined by patchy loss of cytochrome C oxidase (COX) IHC staining within colonic crypts preceded dysplasia and the staining was lowest in regions of dysplasia and adjacent regions. Chronic inflammation in UC results in the generation of reactive oxygen and nitrogen species, leading to the accumulation of DNA damage, which can be measured by 8-nitroguanine (8-NG) and 8-oxo-7,8-dyhydro-2’-deoxyguanosine (8-oxodG). IHC of rectal mucosa from patients with UC-associated dysplasia showed statistically significant increases in 8-NG and decreases in 8-oxodG compared to UC non-neoplastic controls.
Proteomics has also been applied in colitis-associated colon cancer studies with research interests ranging from investigation of disease mechanism to biomarker discovery. To investigate proteomic alterations linked to UC-associated dysplasia and invasive cancer, one study applied stable isotope label based quantitative proteomics to examine the protein expression in the colonic mucosa. The study identified a roster of proteins that displayed at least a 2-fold expression change in random non-dysplastic colon biopsies from Progressors compared to random biopsies from Non-progressors. Among the differentially expressed proteins in the random non-dysplastic biopsies, almost 60% of them were also concurrently expressed in the dysplastic tissues from the same Progressors. These findings suggest that changes in protein expression occur very early in the neoplastic process, before the histologic changes become evident in epithelial cells. Protein activities associated with neoplastic progression included proteins related to mitochondrial function, oxidative activity, and calcium-binding. Protein network analysis suggested that SP1 and c-MYC may play key roles in UC early and late stages of neoplastic progression, respectively.
In a follow-up quantitative proteomics study, individual random rectal samples from UC Progressors were profiled compared to random rectal samples from UC Non-progressors. The study identified over 60 proteins that were differentially expressed in both non-dysplastic rectal tissue and the corresponding dysplastic colonic tissue from Progressors. Mitochondrial proteins, cytoskeletal proteins, RAS superfamily and proteins related to apoptosis were the important protein classes differentially associated with Progressors. One of the mitochondrial proteins, TRAP1, was further validated by IHC in an independent UC cohort, and showed up-regulation in the colon tissues of UC Progressors, but not in the colon tissues of UC Non-progressors. Moreover, up-regulation of TRAP1 preceded the neoplastic changes: it was present in both the dysplastic and non-dysplastic tissue of UC Progressors. TRAP1 staining in dysplastic tissue could achieve 94% sensitivity and 80% specificity in separating Progressors from Non-progressors. In random non-dysplastic rectal biopsies, TRAP1 staining could separate Progressors from Non-progressors with 59% sensitivity and 80% specificity.
Due to the close proximity to intestinal mucosa, stool samples are a rich resource of material for UC biomarker development. It is estimated that approximately one half of stool is comprised of gut microflora and that upwards of one million colon epithelial cells can be isolated from one gram of stool, which would allow for the study of changes in the genetic material or proteome of a patient’s colonocytes, as well as their gut microbiome. Given the complexity of stool samples, it is critical to distinguish changes caused by dysplasia or disease progression, chronic inflammation or consequences of inflammation (i.e., DNA damage, reactive oxygen species, etc.), microbial changes, and/or diet. While biomarkers for UC-associated cancer from stool have not yet been adapted for clinical use, fecal biomarkers are being used to diagnose IBD and irritable bowel syndrome (IBS) patients, to distinguish between UC and CD patients, as well as to assess active inflammation. Currently, the two most widely used stool biomarkers, calprotectin and lactoferrin, are derived from neutrophils, which penetrate the intestinal mucosa in areas of active inflammation and are subsequently shed into the lumen. Both biomarkers may be valuable in assessing the active inflammation as a way to gauge therapeutic response in UC patients and as a non-invasive method to monitor relapses. Use of NMR-based spectroscopy to analyze stool is also gaining appeal due to minimal sample preparation, the ability to detect multiple metabolites at once, and the high reproducibility. Mass resonance metabolomics studies have revealed reductions in buryrate, acetate, methylamine, and trimethylamine as well as differences in the levels of several amino acids when stool samples of UC patients are compared to controls. In addition to protein and metabolite markers, DNA isolated from stool offers a promise as a UC biomarker source. Although mutational analysis failed to discriminate UC-associated cancer samples from cancer-free patients, the methylation changes of four genes - including vimentin, eyes absent homolog 4 (EYA4), bone morphogenetic protein 3 (BMP3), and N-myc downstream regulated gene 4 (NDRG4) - showed AUC (areas under the ROC curve) ranging from 0.84-0.91 in distinguishing UC patients with or without neoplasia, suggesting both high specificity and sensitivity for dysplasia detection.