In this study, we planned to analyze the gene expressions commonly attributed to colonic carcinogenesis from the literature mainly in harvested ACFs compared to their expressions in CRC and normal colonic mucosa. We followed a harvesting protocol that mimics the magnified chromoendoscopic procedure to characterize them further. We found that in both ACF and CRC, multiple genes belonging to different genetic pathways were up-regulated or down-regulated, compared to the normal colonic mucosa, while some of the up-regulated and down-regulated genes in ACF and CRC were in common. Interestingly, the expression of the selected genes was not statistically different between ACF and corresponding CRC in our study. The gene expressions were correlated with tumor size, lymph node metastasis, and tumor laterality, that is, origin in the right or left colon. Protein expression analysis was performed for selected gene products due to the constrain of scant leftover tissue and showed a similar pattern of protein expression as noted with the mRNA mean fold change. This study emphasizes the observation that even tiny, microscopic preneoplastic lesions of the colon as ACFs bear complex molecular alterations like that of a fully developed CRC, in contrast to the conventional belief of their sequential accumulation from low to higher stages.
Many researchers have attempted to study the molecular events in ACF since 1994. However, these studies fail to establish the comprehensive molecular phenotype of ACF. Our data in this study are concordant with the observation by Pretlow TP et al. that all ACF do not harbour APC mutation, rather KRAS mutation is common in them . Mutations in the KRAS gene were the first genetic alteration detected in ACF induced in rodents . According to the literature search, it appears that CDKN2A (p16) and CDKN1A (p21) cyclin-dependent kinase inhibitors are frequently inactivated in a subset of CRC, and overexpression of CDKN2A occurs in the early stage with epigenetic silencing, eventually leading to tumor progression, aberrant expression of CDKN2A and poor prognosis [20, 21]. Our findings corroborated in identifying upregulation of CDKN2A in ACF in comparison to normal colonic mucosa . In previous studies, MSI-associated ACFs were identified to have histologically hyperplastic mucosa and were believed to be the precursor of hyperplastic polyp in a serrated pathway of sporadic CRC [23, 24]. We observed significant downregulation of MLH1 in ACF in comparison to normal controls, with no definite correlation of MLH1 expression between the ACF and corresponding CRC and their histologies.
A variety of inflammatory associated genes have been studied in the pathogenesis of CRC. A study by Peinado et al. suggested that these inflammatory genes are involved in the formation of a premetastatic niche required for organ-specific homing of tumor cells in CRC . Wang et al. showed that the CXCR2 level is elevated in sporadic CRC and active IBD . Foersch et al. stated that blocking growth factors such as VEGFR2 signalling can lead to CRC senescence and is correlated with increased patient survival . Both of our inflammatory associated genes VEGFA (p-value – 0.05) and CCL5 (p-value – 0.02) showed statistically significant alterations in ACF. These results also corroborate the similar pathogenetic hypothesis given by Conteduca et al in 2013, stating that ACF can arise not only from major three molecular pathways but due to the amalgamation of activation of growth factors, stem cell, and chronic inflammation pathways .
In addition, IGF-1, a multifunctional peptide hormone was upregulated in ACF in this study, with its correlation with lymph node metastasis. However, the corresponding CRC included did not show up-regulation of IGF1. This finding contrasted with the findings of Li ZJ et al. and Shiratsuchi I et al [28, 29]. The EGFR gene stimulates the proliferation of neoplastic cells through the MAPK pathway and is associated with an increased pathological tumor stage . Hardwick JC et al. proposed that MAPK is overexpressed and active in colorectal cells by IHC . Although we identified EGFR upregulation in ACF, in the corresponding CRC we did not identify the same, therefore we were unable to comment on its role in colon carcinogenesis. Our study is like the previous studies which don’t show much difference in terms of molecular alteration between ACF and CRC [12, 13, 32].
No definitive correlation was found between topographic pit patterns and ACF histology and the genetic changes observed in this study. In our previous studies [9, 10], although we identified a correlation of the Zyriform pit pattern with ACF of the left colon, dysplastic histology, and TP53 expression, in the index study we did not identify any distinct correlation of topographic or histological ACF types with gene expressions. This may be either due to the low number of successful cases included in this study or it may be possible that all mucosal ACF-like lesions are genetically complex.
This study adds to the fact that ACFs can be sampled and characterized microscopically and genetically. As currently ACF-like lesions are not sampled in vivo by chromoendoscopy for screening purposes, this study was planned in fresh colectomy specimens to harvest and confirm ACF topographically and microscopically. Although FFPE tissue samples have been shown to have lower specificity and sensitivity than fresh frozen samples , they are an important source of material for an exploratory study such as this. Damage to nucleic acid sustained secondary to fixation, storage, and embedding processes can hamper gene expression studies, and we accept this limitation. However, the utility of automated methods adopted in this study showed good specificity (70%) and sensitivity (85%) even in FFPE samples . We also did not perform laser microdissection of the ACF keeping in mind the microscopic nature of these lesions and assuming that the microdissected lesions will not produce a satisfactory amount of RNA for further analyses. We also thought this is not essential, as we assumed similar 'tumor field effects' in the ACF and its immediate adjacent mucosa. We earlier observed that the tumor field effect is observed up to a 5 cm distance of the main tumor mass; therefore, in this study, the macroscopically normal colonic mucosa was sampled from an area at least 5 cm away from the main tumor mass. However, this can be still debated. In addition, as in some of our cases, the RNA yield was low; our successful case numbers for each gene varied. Though performing western blot analysis could have given more quantifiable data, we resorted to the IHC technique keeping in mind the limited left-over tissue in the FFPE blocks already used for sectioning for RNA extraction.
To conclude, ACFs in the human colon show complex activation of pro-oncogenes, cell cycle regulator genes, apoptosis suppressor and activator genes, and inflammation regulatory genes such as the corresponding fully developed CRC. Our observation strengthens the fact that these microscopic lesions can be good candidates for screening and genetic analysis in high-risk patients by high-resolution chromoendoscopy. However, no significant correlation was observed between the topographic and histological characteristics of the ACFs & molecular changes studied; it seems that all ACF-like lesions can be sampled without worrying about their pit patterns. Further multicentre studies should strengthen our observation.