Cimp-Positive Status is More Representative in Multiple Colorectal Cancers than in Unique Primary Colorectal Cancers

Colorectal cancer (CRC) with CpG island methylator phenotype (CIMP) is recognized as a subgroup of CRC that shows association with particular genetic defects and patient outcomes. We analyzed CIMP status of 229 individuals with CRC using an eight-marker panel (CACNA1G, CDKN2A, CRABP1, IGF2, MLH1, NEUROG1, RUNX3 and SOCS1); CIMP-(+) tumors were defined as having ≥ 5 methylated markers. Patients were divided into individuals who developed a “unique” CRC, which were subclassified into early-onset CRC (EOCRC) and late-onset CRC (LOCRC), and patients with multiple primary CRCs subclassified into synchronous CRC (SCRC) and metachronous CRC (MCRC). We found 9 (15.2%) CIMP-(+) EOCRC patients related with the proximal colon (p = 0.008), and 19 (26.8%) CIMP-(+) LOCRC patients associated with tumor differentiation (p = 0.045), MSI status (p = 0.021) and BRAF mutation (p = 0.001). Thirty-five (64.8%) SCRC patients had at least one CIMP-(+) tumor and 20 (44.4%) MCRC patients presented their first tumor as CIMP-(+). Thirty-nine (72.2%) SCRC patients showed concordant CIMP status in their simultaneous tumors. The differences in CIMP-(+) frequency between groups may reflect the importance of taking into account several criteria for the development of multiple primary neoplasms. Additionally, the concordance between synchronous tumors suggests CIMP status is generally maintained in SCRC patients.

pathways are widely accepted to be involved in the etiology of CRC: Chromosomal Instability (CIN), Microsatellite Instability (MSI) and CpG Island Methylator Phenotype (CIMP) [5][6][7] . Structural rearrangements as well as gains and losses of chromosome fragments are characteristic features of CIN tumors, possibly associated with higher mutation rates; the majority of sporadic cases are in this group 8 . MSI is associated with changes in short microsatellite repeats, caused by deficient mismatch repair (MMR) genes, and is related to hereditary non-polyposis colorectal cancer (HNPCC), also called Lynch Syndrome (LS), and some sporadic cases in the elderly population 9 . Lastly, tumor suppressor and DNA repair genes are frequently transcriptionally silenced in CIMP cases. Furthermore, several studies have shown that CIMP positivity is associated with proximal colon location, presence of mucinous features, poor tumor differentiation, MSI, female gender and high BRAF mutation rates 10,11 .
Genetic, biological and clinical differences have been identified depending on the age of onset of CRC in many studies, thus it has been suggested that CRC should be subclassified attending this major criterion [12][13][14][15] . There are other features that should also be taken into account for subclassification, such as the development of two or more different tumors because these cases provide a good model to examine "field effect". This effect is associated with the tendency of healthy colorectal mucosa to suffer early molecular alterations that trigger malignant transformation of the tissue 16,17 . Synchronous CRC (SCRC) is defined by the presence of more than one tumor simultaneously, while metachronous CRC (MCRC) is characterized by the development of a second lesion after surgery and/or diagnosis of the primary tumor [18][19][20] .
Given the above, when focusing on CIMP status, it is important to take the type of CRC, into consideration, since CIMP status affects the response to therapy 21,22 and it may have a relation to the "field effect" linked to CRC 23,24 . For this reason, we examined CIMP status in different CRC subtypes: patients with a single tumor ("unique" CRC), divided into early-onset CRC (EOCRC; age at diagnosis ≤ 45 years old) and patients with late-onset CRC (LOCRC; age at diagnosis > 70 years old); and individuals diagnosed with multiple primary CRC, i.e. patients diagnosed with SCRC or MCRC.

patients.
A total of 229 CRC patients were included in this study at 12 de Octubre Hospital (Madrid, Spain).
Among these patients (all of them Caucasian), 59 had EOCRC, 71 had LOCRC, 54 were diagnosed with SCRC and 45 were diagnosed with MCRC. SCRC was diagnosed when two or more histologically different lesions were developed simultaneously or within a time lapse shorter than six months after the detection of the first tumor 19 . When a secondary neoplasm was detected outside the anastomosis area after more than 6 months of the initial tumor diagnosis, it was considered as MCRC 25 . Samples with the highest content of tumor tissue were selected for molecular analysis. A multi-site database was used to collect clinicopathological, therapeutic and pre-and post-operative information. Informed consent was signed by all patients or by a first degree relative when the patient had died. The protocol of this study was approved by the Ethics Committee of this Institution. DNA isolation. Formalin-fixed paraffin-embedded (FFPE) tumor tissue samples were selected by a pathologist. To be suitable for molecular analysis, one sample should contain a minimum percentage of 70% tumor cells. Briefly, DNA was extracted from FFPE samples using mineral oil to dissolve the paraffin, followed by proteinase K digestion and ethanol precipitation. DNA was further purified and eluted using CLART ® HPV2 kit (Genomica S.A.U., Madrid, Spain).
Microsatellite analysis and analysis of BRAF mutation, hypermethylation of MLH1 and germline mutations in MMR genes. Microsatellite instability status of each tumor was defined using the Bethesda five-marker microsatellite panel (NR-21, BAT-26, BAT-25, NR-24 and MONO-27) 26 . Fluorescence labeled primers were included for amplification of markers, then the PCR products were separated using capillary electrophoresis and analyzed. When 2 or more markers showed instability, the sample was defined as microsatellite instable (Fig. 1). Moreover, MSI tumors were analyzed for the BRAF V600E mutation and hypermethylation of the MLH1 gene promoter in order to verify their sporadic nature, using methylation-specific multiplex ligation-dependent probe amplification (MS-MLPA; ME011-B3, ME0042-CIMP, MRC-Holland, Amsterdam, The Netherlands). MSI cases also were screened for Lynch Syndrome by evaluating germline mutations in the MMR genes (MLH1, MSH2 and MSH6) by high-resolution melting analysis using a LightCycler 480 real-time PCR system (Roche, Mannheim, Germany), as previously reported 27 . KRAs mutations. Mutations in KRAS (codons 12, 13 and 61) were determined previously in patients younger than 45 and older than 70 years. Briefly, DNA was extracted from the neoplastic material, and Sanger sequencing was carried out in both orientations on a 3130 DNA Analyzer (Applied Biosystems, Foster City, CA, USA) 28 . The genetic alterations of KRAS in individuals diagnosed with SCRC and MCRC were evaluated by next generation sequencing using a gene panel related to cancer (Ion PGM System, ThermoFisher, Waltham, MA, USA) (Fig. 2).
CpG island methylator phenotype analysis. For the evaluation of CIMP, we examined the methylation status of the promoter regions of CACNA1G, CDKN2A, CRABP1, IGF2, MLH1, NEUROG1, RUNX3 and SOCS1 using the SALSA MS-MLPA Probemix (ME0042-CIMP, MRC-Holland, Amsterdam, The Netherlands). Each patient was classified as CIMP-(+) or CIMP-(−) depending on whether tumors showed ≥ 5/8 or < 5/8 methylated promoters, respectively (Fig. 3A,B) 11 . Patients diagnosed with multiple CRC and distinct CIMP statuses in their tumors were categorized as CIMP-MM (mismatching). statistical analysis. Clinical characteristics were compared between different groups according to CIMP status, including age, sex, stage, tumor location, presence of mucinous features, BRAF and KRAS mutations, and MSI status. Categorical variables were expressed as number of cases and their percentage, and continuous variables were expressed as mean values plus/minus standard deviation (SD). Comparison of categorical variables was done using Pearson's Chi Square (X 2 ) test. For comparisons between two groups Student's t test was performed, www.nature.com/scientificreports www.nature.com/scientificreports/ and for comparisons between more than two groups analysis of variance (ANOVA) (for normal distributions) or the Kruskal-Wallis test (for nonparametric distributions) were used. Statistical analysis was carried out using IBM SPSS 17.0 (SPSS Inc., Chicago, IL, USA). P value of <0.05 was considered statistically significant differences. ethics approval and consent to participate. Study approval was obtained from the Ethics Committee of the 12 de Octubre University Hospital in Madrid, Spain. All procedures were performed in accordance with the ethical standards of the institutional research committee and the Declaration of Helsinki.

Results
Global features of "unique" colorectal cancer. We collected clinical characteristics of 130 individuals with "unique" CRC, of which 59 were EOCRC cases and the remaining 71 patients were LOCRC cases. The average age at diagnosis was 40 ± 5 and 78 ± 6 years old for EOCRC and LOCRC, respectively. Male to female ratio was >1 in the EOCRC group, whereas in the LOCRC set it was < 1. When we assessed the anatomical location of the tumors, the EOCRC group showed most tumors (47.5%) at the distal colon, while LOCRC patients presented more rectum location (42.3%). Among the 59 patients with EOCRC, 8 (13.6%) patients showed MSI status, of  Global features of primary multiple colorectal cancer. Among 99 cases with multiple CRC, 54 individuals were diagnosed with SCRC and 45 patients with MCRC. In both groups, the mean age at onset was around 70 years old, with a male to female ratio > 1. Moreover, the most common tumor location for both groups was the entire colon, defined as the location of the synchronous or metachronous tumors at different sides of the colon. Only one patient in the SRCR group (1.8%) had MSI in both tumors whereas discordant MSI status was found between synchronous tumors in 3 cases (5.6%), of which 1 (25.0%) case was diagnosed with LS. Five www.nature.com/scientificreports www.nature.com/scientificreports/ (10.4%) and 31 (64.6%) patients presented BRAF mutations and KRAS mutations in at least one tumor, respectively. In the MCRC group, several remarkable features were the diagnosis of metachronous neoplasm at an early stage (82.2%), the total concordance of MSI status between paired tumors, and only 1 (2.2%) patient who was identified as a LS case, showed MSI status. Regarding a BRAF mutation, MCRC patients showed concordance between both tumors, of which 1 (2.2%) patient presented a BRAF mutation. Eighteen (40.0%) MCRC patients showed a KRAS mutation in at least one tumor and 11 (24.4%) in paired-tumors (Tables 3 and 4).  www.nature.com/scientificreports www.nature.com/scientificreports/ second tumor CIMP-(−), and 9 (20.0%) were first tumor CIMP-(−) and second tumor CIMP-(+) (Tables 3 and  4). Furthermore, the concordance of CIMP status between paired tumors in the SCRC and MCRC groups is summarized in Tables 5 and 6. Thirty-nine (72.2%) SCRC patients showed the same CIMP status in both tumors. In the cohort of MCRC, 27 (60%) patients presented concordant CIMP status in both tumors.

Correlation between CIMP phenotype and clinical features.
Regarding clinical correlation with CIMP status, CIMP-(+) tumors in the EOCRC group were mainly located in proximal colon (p = 0.008), with a very low rate of well-differentiated tumors ( Table 1). The CIMP-(+) tumors of patients with LOCRC presented moderate tumor differentiation (p = 0.045), MSI status (p = 0.021) and a higher BRAF mutation rate (p = 0.001) ( Table 2). Other interesting features were the higher proportion of females as well as the rectal tumor location (both 63%). In the SCRC subgroup (Table 3), cases with both tumors CIMP-(+), 16 (80.0%) patients were male and 11 (55.0%) tumors were distal-sided both paired tumors. Most patients diagnosed with MCRC and CIMP-(+) for both tumors showed entire colon location (81.8%), mucin production in the first tumor (25%) and early stage at diagnosis of the second tumor (72.7%) ( Table 4). Differences in diagnostic tumor stage, sex, age and other clinicopathological features analyzed between different and concordant CIMP status in the SCRC and MCRC groups were not statistically significant (Tables 3 and 4).

Discussion
CpG-island methylation testing has been proposed as a tool for cancer detection and prognosis, and indicates that methylation status is of clinical relevance 29 , despite the timing of its occurrence and its interaction with other genetic defects are not fully understood [30][31][32] . In this study, we assessed CIMP status in different subsets of CRCs according to age of onset and the number of primary neoplasms: EOCRC and LOCRC with "unique" CRCs, and SCRC and MCRC.
In the assessment of CIMP status 11,33 , we found that only 15.2% of EOCRC patients and 26.8% of LOCRC patients showed CIMP-(+) tumors. However, 37.0% of SCRC patients were CIMP-(+) for both tumors and   www.nature.com/scientificreports www.nature.com/scientificreports/ 27.8% of the tumor pairs showed at least one CIMP-(+) tumor, suggesting that the serrated pathway of carcinogenesis could be the main mechanism for SCRC development 34,35 . In the MCRC group, 24.4% were CIMP-(+) for both tumors and 40.0% were CIMP-(+) for at least one tumor. The reported differences in CIMP-(+) frequency between different groups could reflect the relative involvement of this particular molecular phenotype throughout the development of multiple neoplasms. In summary, our findings show a higher CIMP-(+) frequency in patients who develop multiple tumors than in patients who develop a single tumor. These different epigenetic patterns may be due to a "field effect" and possibly a highly-susceptible tissue microenvironment could be generated by the interplay of different etiologic factors including lifestyle, eating habits or environmental issues leading to the appearance and development of malignant tumors 20,36 .
Moreover, in this study we observed that most pairs of tumors had concordant CIMP status in patients with SCRC (72.2%) which suggests that CIMP is maintained throughout neoplasm development; this has been taken as a evidence for the high probability that synchronous tumors could be developed through the same genetic pathway in each particular patient 37 . On the other hand, 40% of the patients diagnosed with MCRC did not present concordance concerning CIMP status. The clinical implication of this finding is that the analysis of the CIMP status in any of the synchronous tumors could provide a reasonably reliable prediction of CIMP status in the other tumors even when it is not directly assayed. However, this prediction would be less reliable in MCRC. Comparison of these epigenetic patterns in synchronous and metachronous lesions may suggest that there is a tendency to have clonal features or a stronger "field effect" in patients diagnosed with SCRC, while the tumor heterogeneity in patients with MCRC may be caused by the contribution of different carcinogenic pathways in tumors developed at different time points.
About the results from the correlation between CIMP status and the clinical features within each group of tumors, recent systematic reviews have confirmed the association between CIMP phenotype and older ages, female gender, proximal tumor location, mucinous histology, poor differentiation and MSI 36,37 . According to the "unique" colorectal cancers, EOCRC subset showed the proximal location, mucinous and poorly-differentiated   www.nature.com/scientificreports www.nature.com/scientificreports/ phenotype, while LOCRC linked with a higher proportion within this subset, and gender and MSI phenotype. Maybe the age-of-onset criterion gives rise to this stratification in the characteristics related to the CIMP in this type of CRC. Nevertheless, this characteristics didn´t correlate with Multiple Primary CRCs.
Our results seem to confirm the fact that there are distinct groups of CRC patients. This study, focused on the current knowledge of epigenetic alterations in CRC, could represent a substantial contribution to this research line, since it may have implications in terms of prevention, diagnosis and therapy.

Conclusion
Our results underscore the importance of taking into account several criteria for the development of multiple primary tumors when analyzing CRC. There is higher CIMP-(+) frequency in patients diagnosed with multiple CRC than in patients with "unique" CRC. Additionally, we conclude that there is a concordance of CIMP status of synchronous tumors in SCRC. Therefore, it could be suggested that CIMP status in one of the simultaneous tumors could predict CIMP status of other tumors.

Data Availability
The data that support the findings of this study are available from the corresponding author on reasonable request.