Identifying the Precursor of Colorectal Cancer
Identifying the Precursor of Colorectal Cancer
SSAs were previously reported to be precursors of CRCs with MSI. Although identification of SSAs in screening colonoscopies has important implications for the prevention and early detection of CRCs, the colonoscopic findings for SSAs have been less than definitive. Criteria for colonoscopic diagnosis of colorectal lesions were established on the basis of histological findings; however, there is no clear histological definition to distinguish SSAs from HPs, and the rate of concordance among gastrointestinal pathologists for diagnosis of SSA is low. Detection of SSAs also reportedly differs among endoscopists, and classification of HP and SSA differs among pathologists.
In the present study, we performed an integrated analysis of the genetic, epigenetic and clinical features of serrated lesions in an effort to establish more definitive criteria for the colonoscopic diagnosis of SSA. We identified an SSA-specific pit pattern Type II-O and prospectively validated its clinical utility for detecting SSA. Molecular dissection of mixed serrated lesions suggested that subcomponents with advanced pit patterns were derived from coexisting Type II-O pit-positive SSAs, and additional molecular alterations (e.g., p16 and MLH1 methylation) were acquired during tumorigenesis. The inactivation of cell cycle regulatory genes, including p16 and p53, has an important role in CRC development by enabling tumor cells to escape oncogene-induced cellular senescence.IGFBP7 has been shown to have a central role in oncogenic BRAF-induced senescence, and it is also a direct target of p53. In CRCs, methylation of IGFBP7 is strongly associated with BRAF mutation, the lack of p53 mutation and the presence of MSI and CIMP. In the present study, we showed that levels of IGFBP7 methylation are elevated in Type II-O pit-positive serrated lesions, suggesting the tumor suppressor function of p53 is attenuated at an early stage of the MSI pathway.
The mechanism underlying the strong relationship between Type II-O pits and SSAs remains unclear. SSAs exhibit a variety of histological features, including exaggerated serration, boot-shaped crypts, and the branching and dilatation of the crypts. These features are usually observed near the base of the crypts, making it difficult to endoscopically discriminate SSAs from HPs. SSAs are often covered by abundant mucus production, and accumulation of the mucin within crypts may lead to their dilatation. It is thus conceivable that overproduction of mucin may be associated with the Type II-O pit pattern in SSAs.
Identification of SSAs in screening colonoscopies has great significance for the prevention and surveillance of CRCs, as it is generally accepted that colonoscopy and polypectomy reduce the incidence of CRCs. Although SSAs are usually treated as conventional adenomas, it is still uncertain whether all SSAs should be endoscopically resected. In the present study, a large majority of Type II-O-positive lesions were histologically SSAs, and were tightly associated with BRAF mutation and CIMP, suggesting these lesions are appropriate targets for endoscopic resection. Moreover, as serrated lesions with Type II-O plus Type III, IV or V pits exhibit additional malignant potential, they too should be targets of early treatment. On the other hand, we noted that approximately half of the serrated lesions with only conventional Type II pits in our validation set were diagnosed as SSAs, though all Type II lesions in the training set were HP/IMs. This difference reflects the extremely high specificity and relatively low sensitivity of Type II-O pits for distinguishing SSAs from HP/IMs (Table 1).
There are several possible explanations for the low sensitivity of Type II-O pits for definition of SSAs. First, SSAs with only conventional Type II pits in our validation set may have been at a very early stage of development, where Type II-O pits were not yet established. Alternatively, it may simply reflect the difficulty of histologically distinguishing SSAs from HP/IMs. Of these two possibilities, the first is supported by the fact that although BRAF was frequently mutated in these lesions, aberrant DNA methylation was infrequent and only a limited number of the specimens were CIMP-positive. Further prospective study will be needed to unravel the time course of the emergence of Type II-O pits in the serrated pathway.
There are several potential limitations to the current study. First, our diagnostic system is based on endoscopic observations, so it may be affected by the skills of the endoscopists. As described above, among the ten endoscopists participating in this study, four had experience with more than 1,000 endoscopy cases, whereas the remaining six were less experienced. We therefore divided the data into two parts according the endoscopists' experience and compared the diagnostic accuracy of Type II-O pits for defining SSAs. As shown in Supplementary Table 4, the sensitivities and specificities were similar between the two groups, suggesting the intra-observer variability in our diagnostic system is relatively limited. Second, there are several biases in the selection of tumor specimens in this study. For instance, the average size of the lesions is relatively large (>10 mm). This is because these specimens were collected through endoscopic resection, and smaller lesions, which are usually subject to follow-up observation, were less likely to be included in this study. In addition, during the collection of the training set specimens, we treated as many possible serrated lesions as possible, which might have resulted in a somewhat high frequency of TSAs in our training set. Moreover, the identification of Type II-O pits enabled us to distinguish non-neoplastic HP/IM from neoplastic serrated lesions, which might have reduced the frequency of HP/IM in the validation set. Thus, our findings should be validated in an independent multicenter study that includes a larger number of samples. Third, our study does not provide effective criteria to endoscopically define TSAs. Further study may enable us to find new clues to refine the endoscopic diagnosis of serrated lesions.
In summary, we have identified a novel surface microstructure that is specific to SSAs. Recent studies have shown that the presence of SSAs is associated with an increased risk of synchronous CRCs. In the context of those findings, our observations indicate detection of Type II-O pits could be predictive of CRC risk. Thus, more intensive and frequent colonoscopic surveillance may be appropriate for patients in whom Type II-O-positive lesions were once detected. Not only could these findings contribute to the prevention of MSI-positive CRCs, they also demonstrate that integrative analysis of molecular and endoscopic characteristics greatly improves our understanding of the pathogenesis of CRC and the quality of colonoscopic surveillance.
Discussion
SSAs were previously reported to be precursors of CRCs with MSI. Although identification of SSAs in screening colonoscopies has important implications for the prevention and early detection of CRCs, the colonoscopic findings for SSAs have been less than definitive. Criteria for colonoscopic diagnosis of colorectal lesions were established on the basis of histological findings; however, there is no clear histological definition to distinguish SSAs from HPs, and the rate of concordance among gastrointestinal pathologists for diagnosis of SSA is low. Detection of SSAs also reportedly differs among endoscopists, and classification of HP and SSA differs among pathologists.
In the present study, we performed an integrated analysis of the genetic, epigenetic and clinical features of serrated lesions in an effort to establish more definitive criteria for the colonoscopic diagnosis of SSA. We identified an SSA-specific pit pattern Type II-O and prospectively validated its clinical utility for detecting SSA. Molecular dissection of mixed serrated lesions suggested that subcomponents with advanced pit patterns were derived from coexisting Type II-O pit-positive SSAs, and additional molecular alterations (e.g., p16 and MLH1 methylation) were acquired during tumorigenesis. The inactivation of cell cycle regulatory genes, including p16 and p53, has an important role in CRC development by enabling tumor cells to escape oncogene-induced cellular senescence.IGFBP7 has been shown to have a central role in oncogenic BRAF-induced senescence, and it is also a direct target of p53. In CRCs, methylation of IGFBP7 is strongly associated with BRAF mutation, the lack of p53 mutation and the presence of MSI and CIMP. In the present study, we showed that levels of IGFBP7 methylation are elevated in Type II-O pit-positive serrated lesions, suggesting the tumor suppressor function of p53 is attenuated at an early stage of the MSI pathway.
The mechanism underlying the strong relationship between Type II-O pits and SSAs remains unclear. SSAs exhibit a variety of histological features, including exaggerated serration, boot-shaped crypts, and the branching and dilatation of the crypts. These features are usually observed near the base of the crypts, making it difficult to endoscopically discriminate SSAs from HPs. SSAs are often covered by abundant mucus production, and accumulation of the mucin within crypts may lead to their dilatation. It is thus conceivable that overproduction of mucin may be associated with the Type II-O pit pattern in SSAs.
Identification of SSAs in screening colonoscopies has great significance for the prevention and surveillance of CRCs, as it is generally accepted that colonoscopy and polypectomy reduce the incidence of CRCs. Although SSAs are usually treated as conventional adenomas, it is still uncertain whether all SSAs should be endoscopically resected. In the present study, a large majority of Type II-O-positive lesions were histologically SSAs, and were tightly associated with BRAF mutation and CIMP, suggesting these lesions are appropriate targets for endoscopic resection. Moreover, as serrated lesions with Type II-O plus Type III, IV or V pits exhibit additional malignant potential, they too should be targets of early treatment. On the other hand, we noted that approximately half of the serrated lesions with only conventional Type II pits in our validation set were diagnosed as SSAs, though all Type II lesions in the training set were HP/IMs. This difference reflects the extremely high specificity and relatively low sensitivity of Type II-O pits for distinguishing SSAs from HP/IMs (Table 1).
There are several possible explanations for the low sensitivity of Type II-O pits for definition of SSAs. First, SSAs with only conventional Type II pits in our validation set may have been at a very early stage of development, where Type II-O pits were not yet established. Alternatively, it may simply reflect the difficulty of histologically distinguishing SSAs from HP/IMs. Of these two possibilities, the first is supported by the fact that although BRAF was frequently mutated in these lesions, aberrant DNA methylation was infrequent and only a limited number of the specimens were CIMP-positive. Further prospective study will be needed to unravel the time course of the emergence of Type II-O pits in the serrated pathway.
There are several potential limitations to the current study. First, our diagnostic system is based on endoscopic observations, so it may be affected by the skills of the endoscopists. As described above, among the ten endoscopists participating in this study, four had experience with more than 1,000 endoscopy cases, whereas the remaining six were less experienced. We therefore divided the data into two parts according the endoscopists' experience and compared the diagnostic accuracy of Type II-O pits for defining SSAs. As shown in Supplementary Table 4, the sensitivities and specificities were similar between the two groups, suggesting the intra-observer variability in our diagnostic system is relatively limited. Second, there are several biases in the selection of tumor specimens in this study. For instance, the average size of the lesions is relatively large (>10 mm). This is because these specimens were collected through endoscopic resection, and smaller lesions, which are usually subject to follow-up observation, were less likely to be included in this study. In addition, during the collection of the training set specimens, we treated as many possible serrated lesions as possible, which might have resulted in a somewhat high frequency of TSAs in our training set. Moreover, the identification of Type II-O pits enabled us to distinguish non-neoplastic HP/IM from neoplastic serrated lesions, which might have reduced the frequency of HP/IM in the validation set. Thus, our findings should be validated in an independent multicenter study that includes a larger number of samples. Third, our study does not provide effective criteria to endoscopically define TSAs. Further study may enable us to find new clues to refine the endoscopic diagnosis of serrated lesions.
In summary, we have identified a novel surface microstructure that is specific to SSAs. Recent studies have shown that the presence of SSAs is associated with an increased risk of synchronous CRCs. In the context of those findings, our observations indicate detection of Type II-O pits could be predictive of CRC risk. Thus, more intensive and frequent colonoscopic surveillance may be appropriate for patients in whom Type II-O-positive lesions were once detected. Not only could these findings contribute to the prevention of MSI-positive CRCs, they also demonstrate that integrative analysis of molecular and endoscopic characteristics greatly improves our understanding of the pathogenesis of CRC and the quality of colonoscopic surveillance.