|Home | About | Journals | Submit | Contact Us | Français|
Synchronous colorectal neoplasias (2 or more primary carcinomas identified in the same patient) are caused by common genetic and environmental factors and can therefore be used to study the field effect. Synchronous colon cancers have not been compared with control solitary cancers in a prospective study.
We analyzed data collected from 47 patients with synchronous colorectal cancers and 2021 solitary colorectal cancers (controls) in 2 prospective cohort studies. Tumors samples were analyzed for methylation in LINE-1 and 16 CpG islands (CACNA1G, CDKN2A [p16], CRABP1, IGF2, MLH1, NEUROG1, RUNX3, SOCS1, CHFR,HIC1, IGFBP3, MGMT, MINT1, MINT31, p14 [ARF], and WRN); microsatellite instability (MSI); the CpG island methylator phenotype (CIMP); 18q loss of heterozygosity; KRAS, BRAF and PIK3CA mutations; and expression of β-catenin, p53, p21, p27, cyclin D1, fatty acid synthase, and cyclooxygenase-2.
Compared to patients with solitary colorectal cancer, synchronous colorectal cancer patients had reduced overall survival time (log-rank p=0.0048; hazard ratio [HR]=1.71; 95% confidence interval [CI]=1.17–2.50; p=0.0053; multivariate HR=1.47; 95% CI=1.00–2.17; p=0.049). Compared to solitary tumors, synchronous tumors more frequently contained BRAF mutations (p=0.0041), CIMP-high (≥6/8 methylated CIMP markers; p=0.013), and MSI-high (p=0.037). Methylation levels of LINE-1 (Spearman r=0.82; p=0.0072) and levels of CpG island methylation (p<0.0001) correlated between synchronous cancer pairs from the same individuals.
Synchronous colorectal cancers had more frequent mutations in BRAF, were more frequently categorized as CIMP- and MSI-high, and a worse prognosis than solitary colorectal cancers. Similar epigenomic and epigenetic events were frequently observed within a synchronous cancer pair, suggesting the presence of a field effect.
Synchronous colorectal cancers refer to two or more primary colorectal carcinomas detected in a single individual at the time of the first diagnosis of colorectal cancer. Synchronous neoplasias, which arise in a background of common etiologic (genetic or environmental) factors, can provide a unique model to examine molecular aberrations and field effect.1 Random molecular aberrations within synchronous tumor pairs may support a stochastic process in carcinogenesis, while nonrandom molecular aberrations may support a specific etiology or cause in carcinogenic process,1 or the hypothesis of field effect in apparently normal colonic mucosa.2–4 Previous studies have analyzed several molecular markers [p53,5 microsatellite instability (MSI),6–10 MLH1,6, 7, 9–11 MSH2,6, 7, 9, 11MSH6,7 or CpG island methylation1] within synchronous colorectal cancer pairs, suggesting both concordant1, 8–10 and discordant1, 5, 7, 10, 11 alteration patterns. However, no previous study has examined global DNA methylation levels in synchronous colorectal cancer pairs. Global DNA hypomethylation has been linked to genomic instability and carcinogenesis,12, 13 and hypomethylation in LINE-1 repetitive sequence has been associated with poor prognosis in colon cancer.14
Previous studies have examined clinical features of synchronous colorectal cancer patients (Table 1; individual studies are listed in Supplemental Table 1).5–11, 15–26 However, the prognostic significance of cancer synchronicity remains inconclusive.17–24 While synchronous cancer patients and solitary cancer patients showed similar survival in most studies,17, 19–24 one study showed worse survival in synchronous cases.18 However, in all of these studies,17–24 “control” solitary cancers were retrospectively selected, thereby subject to potential selection bias. An optimal control group would be solitary colorectal cancers in a population that has given rise to synchronous colorectal cancers. It is possible to secure such a control group in a prospective cohort setting.
In this study, during follow-up of two well-characterized, prospective cohort studies, we identified 47 cases of synchronous colorectal cancers and 2068 control solitary colorectal cancers that had arisen in the same background population as synchronous cases. We examined patient survival and various molecular changes in synchronous and solitary colorectal cancers in our prospective cohort studies.
We utilized two prospective cohort studies; the Nurses’ Health Study (121,701 women followed since 1976),27 and the Health Professionals Follow-up Study (51,529 men followed since 1986).27 Every 2 years, participants have been sent follow-up questionnaires to identify newly diagnosed cancer and other diseases in themselves and their first degree relatives. For nonresponders, we searched the National Death Index to discover deaths and ascertain any diagnosis of colorectal cancer that contributed to death or was a secondary diagnosis. Study physicians and pathologists reviewed medical and pathology records, and recorded tumor stage, location and synchronicity status.
During prospective follow-up of the cohort participants up to 2002, there were 2068 incident colorectal cancer patients with available pathology reports and follow-up data, which constituted the base of this study (Figure 1). Among them, we identified 47 patients who had synchronous colorectal cancers, which were strictly defined as the presence of two or more colorectal cancers (with at least submucosal invasion, stage pT1) that were grossly, unequivocally separated by normal colorectal mucosa at the first diagnosis of colorectal cancer. In addition, metastasis mimicking synchronous tumors was excluded. Two of the 47 synchronous cases had 3 synchronous cancers, and all of the other cases had 2 synchronous cancers: cancer tissues of the patients with 3 tumors were unavailable. The remaining 2021 patients had solitary colorectal cancers at the first diagnosis, which constituted a control group in this study. These solitary colorectal cancers had arisen in the population that had given rise to synchronous colorectal cancer cases, and thus constituted an optimal comparison (control) group (Table 1). Patients were observed until death or June 2008, whichever came first. We collected paraffin-embedded tissue blocks from hospitals where patients underwent tumor resections,27 and pathologic features were examined by a pathologist (S.O.) unaware of other data. Tumor grade was classified as low or high (<50% or ≥50% solid areas, respectively). Based on tissue specimen availability, we performed pathologic and molecular analysis on a total of 29 cases of synchronous colorectal cancers and 1084 solitary colorectal cancers for comparison. Informed consent was obtained from all 2068 subjects. This study was approved by the Human Subjects Committees at Brigham and Women’s Hospital and the Harvard School of Public Health.
DNA from paraffin-embedded tissue was extracted, and PCR and Pyrosequencing targeted for KRAS codons 12 and 13,28BRAF codon 600,29 and PIK3CA exons 9 and 20 were performed.30 MSI status was determined using D2S123, D5S346, D17S250, BAT25, BAT26, BAT40, D18S55, D18S56, D18S67 and D18S487.31 MSI-high was defined as the presence of instability in ≥30% of the markers, MSI-low as instability in 1–29% of the markers, and microsatellite stability (MSS) as the absence of instability. For 18q loss of heterozygosity (LOH) analysis using microsatellite markers (D18S55, D18S56, D18S67, D18S487), LOH at each locus was defined as ≥40% reduction of one of two allele peaks in tumor DNA relative to normal DNA.32 18q LOH negativity was defined as the presence of ≥2 informative markers and the absence of LOH.
Bisulfite DNA treatment and real-time PCR (MethyLight) assays were validated and performed.33 We quantified methylation at 8 CIMP-specific CpG islands [CACNA1G, CDKN2A (p16), CRABP1, IGF2, MLH1, NEUROG1, RUNX3 and SOCS1].34, 35 CIMP-high was defined as ≥6/8 methylated promoters using the 8-marker CIMP panel, CIMP-low as 1/8–5/8 methylated promoters and CIMP-0 as 0/8 methylated promoters, according to the previously established criteria.34 We also quantified methylation at 8 other loci (CHFR, HIC1, IGFBP3, MGMT, MINT1, MINT31, p14, and WRN).36, 37 The percentage of methylated reference (PMR, i.e., methylation index) at a specific locus was calculated as previously described.33 Methylation positivity was set as PMR ≥4 as previously validated.33 LINE-1 methylation level was measured by Pyrosequencing,38
We constructed tissue microarrays (TMAs).27 Methods of immunohistochemistry were previously described as follows: cyclin D1,39β-catenin,40 p21, p27, p53,41 fatty acid synthase (FASN) and COX-2.42 All immunohistochemically-stained slides for each marker were interpreted by one of the investigators (p53, p21, p27, COX-2 and FASN by S.O.; cyclin D1 and β-catenin by K.N.) unaware of other data. A random sample of 108–402 tumors were reexamined by a second observer (p21, p27 and cyclin D1 by K.S.; p53, FASN by K.N.; β-catenin by S.O.; COX-2 by R. Dehari, Kanagawa Cancer Center) unaware of other data. The concordance rates between the two observers were 0.83 or greater (all κ≥0.60, except for κ=0.57 for FASN; all p<0.0001), indicating good to substantial agreement.
We used SAS version 9.1 (SAS Institute, Cary, NC) and all p values were two-sided. Rigorous corrections for multiple hypothesis testing were not attempted due to limited statistical power. The χ2 test (or Fisher’s exact test when any expected cell count was <5) was used to examine an association between categorical variables. The Wilcoxon rank-sum test was performed to compare age and LINE-1 methylation levels between synchronous cancers and solitary controls. Spearman correlation coefficients were used to assess the correlation of LINE-1 methylation levels within synchronous colorectal cancer pairs. The κ coefficient was calculated to assess agreement within synchronous cancer pairs on tumor location (proximal colon vs. distal colorectum) and methylation status (positive vs. negative) at the 16 individual CpG islands.
The Kaplan-Meier method was used to describe the distribution of survival time, and the log-rank test was performed. For analyses of colorectal cancer-specific mortality, death as a result of colorectal cancer was the primary end point and deaths from other causes were censored. Stage-matched (stratified) Cox proportional hazard models were used to calculate mortality by tumor synchronicity status, adjusted for age at diagnosis (continuous), sex, year of diagnosis (continuous), body mass index (BMI; <30 vs. ≥30 kg/m2) prior to diagnosis, family history of colorectal cancer in any first degree relative (present vs. absent), tumor location (proximal vs. distal colon vs. rectum). Tumor stage (I, II, III, IV, unknown) was used as a matching (stratifying) variable using the “strata” option in the SAS “proc phreg” command, to avoid residual confounding and overfitting. For cases with missing data in BMI (14% missing) or tumor location (2.6% missing), we included those cases in the category of BMI <30kg/m2, or a category of “proximal colon”, respectively, in order to avoid overfitting. Neither BMI nor tumor location was a confounder. We confirmed that excluding cases with missing BMI data (or tumor location data) did not substantially alter results (data not shown).
During follow-up of the two prospective cohort studies (the Nurses’ Health Study and the Health Professionals Follow-up Study), there were 2068 eligible colorectal cancer patients diagnosed up to 2002. Among the 2068 patients, we identified 47 cases of synchronous colorectal cancers (Figure 1). The remaining 2021 solitary colorectal cancers had arisen in the same cohort population that had given rise to the 47 synchronous cases, and thus constituted an optimal control group. We assessed clinical features of the 47 synchronous colorectal cancer cases and the 2021 control solitary cancer controls (Table 2). Compared to the solitary cancer patients, the synchronous cases were significantly older (p=0.016). Synchronous cancers within pairs tended to co-localize in either proximal colon or distal colorectum (71% concordant; κ=0.42; p=0.0057) (Supplemental Table 2).
We performed survival analysis on the 2068 patients with incident colorectal cancers. During follow-up, there were 875 deaths, including 579 colorectal cancer-specific deaths. In Kaplan-Meier analysis, synchronous cancer patients experienced worse survival than solitary cancer patients (log-rank p=0.0048) (Figure 2). In a univariate Cox model, compared to solitary cancer patients, synchronous cancer patients experienced significantly higher overall mortality [hazard ratio (HR) 1.71; 95% confidence interval (CI), 1.17–2.50; p=0.0053] (Table 3). In a multivariate Cox model adjusting for potential confounders, synchronous cases were associated with higher overall mortality (adjusted HR 1.47; 95% CI, 1.00–2.17; p=0.049). Similar results were obtained when colorectal cancer-specific mortality was used as the endpoint, though statistical significance was not reached in adjusted models. No major confounder was identified in the Cox regression analyses.
Among the 2068 colorectal cancer patients, tumor tissues were available for pathologic and molecular analyses in 1113 patients, including 29 synchronous cases for which tissue was obtained from at least one tumor (a tumor with higher stage or larger tumor if synchronous tumors were the same stage). We compared pathologic and molecular characteristics of synchronous cancers with those of solitary cancer controls (Table 2). Compared to the solitary cancers, synchronous cancer cases more frequently showed BRAF mutations (p=0.0041), CpG island methylator phenotype (CIMP)-high (p=0.013) and microsatellite instability (MSI)-high (p=0.037). Synchronous cases were associated with hypermethylation at SOCS1 (p=0.042) among the 16 CpG islands (Supplemental Table 3).
The association between synchronicity and BRAF mutation might have been due to confounding by tumor location (i.e., proximal colon) or age. Thus, we examined the relation between synchronicity and BRAF mutation in the proximal colon. In the proximal colon, BRAF mutations were significantly more common in synchronous cancers (58%=7/12; p=0.009) than in solitary cancers (22%=79/353). We also constructed a logistic regression model including age, BRAF mutation, and synchronicity status as an outcome variable. The association between BRAF and tumor synchronicity persisted (adjusted odds ratio, 4.02; 95% CI, 1.66–9.72; p=0.0021), while age was no longer associated with tumor synchronicity (p=0.94).
Since BRAF mutation is closely associated with CIMP-high in colorectal cancer,43 we examined the frequencies of combined BRAF/CIMP subtypes. Synchronous cancers were more commonly BRAF-mutated/CIMP-high (32%=7/22), compared to solitary controls (8.0%=63/787; p=0.0058) (Table 2).
The relation between tumor synchronicity and high mortality could be attributed to molecular features of synchronous tumors. Thus, we limited our cases to patients with available tumor tissue data and adjusted for tumor stage and molecular variables including BRAF mutation,36, 44 CIMP,36 MSI,45 and tumor grade. The adjusted HR for overall mortality (synchronous cases vs. solitary controls) was 1.92 (95% CI, 1.10–3.35; p=0.022), suggesting that the inferior survival associated with synchronous cancers was independent of BRAF mutation, CIMP, MSI and tumor grade.
Among the 29 synchronous cancer cases with available tumor tissue, 10 synchronous cancer cases could provide cancer tissues from both cancers. Among the 10 cases in Table 4, our previous study1 examined Case Nos. 1–6 for methylation in 6 CpG islands (CACNA1G, CDKN2A, CRABP1, MLH1, NEUROG1 and MGMT), MSI, KRAS and BRAF. In this study, we considerably expanded analyses in terms of the number of markers and sample size, and we examined adjacent normal colon for CIMP, LINE-1 and BRAF mutation. Notably, LINE-1 methylation levels in the synchronous cancer pairs were significantly correlated (Spearman r = 0.82; p=0.0072) (Figure 3). This correlation appeared to persist even in the tumor pairs of discordant anatomical locations (i.e., one proximal tumor and the other distal tumor). LINE-1 methylation level in adjacent normal colon (from 7 cancer pairs) was consistently high (>71%) even in the cases with low LINE-1 methylation level in both synchronous tumors (Case Nos. 5, 9 and 10; Supplemental Table 4).
With regard to CpG island methylation, Table 5 summarizes pairwise agreements within the 10 synchronous cancer pairs on methylation status at each of the 16 CpG islands. Within all of the 10 synchronous cancer pairs, methylation status at the 16 CpG islands were significantly concordant (78% concordant, κ=0.53, p<0.0001). In addition, synchronous cancer pairs of concordant anatomical locations (i.e., proximal colon vs. distal colorectum) showed a significant concordance (81% concordant, κ=0.41, p=0.010, for proximal pairs; 94% concordant, κ=0.80, p=0.0001, for distal pairs). In contrast, synchronous pairs in discordant locations (i.e., one proximal cancer and the other distal cancer) did not show concordant methylation (κ=0.15, p=0.18).
Finally, we analyzed BRAF mutation and DNA methylation at the 8 CIMP-specific CpG islands in adjacent normal colon in some synchronous cases, and neither BRAF mutation nor substantial CpG island methylation was detected in normal colon (Supplemental Table 4).
In this study, we examined genetic and epigenetic features as well as clinical outcome of synchronous colorectal cancer cases. To date, no previous study has examined global DNA methylation levels within synchronous colorectal cancer pairs. In addition, no previous study has compared clinical and molecular features of synchronous colorectal cancers with those of solitary colorectal cancers which had occurred in the same background population as the synchronous cancers in a prospective cohort setting (Table 1). We found that synchronous colorectal cancer patients experienced a significantly higher overall mortality than solitary cancer patients. Compared to solitary cancers, synchronous colorectal cancers more commonly showed BRAF mutations, CIMP-high and MSI-high. In our pairwise study, synchronous colorectal cancer pairs exhibited a significant correlation of LINE-1 methylation levels. Synchronous cancer pairs in proximal colon and those in distal colorectum showed a significantly concordant pattern of CpG island methylation, while tumor pairs in discordant locations (one proximal cancer and the other distal cancer) did not show concordant CpG island methylation. Our data indicate that synchronous colorectal cancers give us unique insights into field effect (cancerization) and colorectal carcinogenic process.
The proportion of synchronous cancer cases among colorectal cancer patients vary from 1.1% to 5.3%.5, 9, 10, 16–26 The relatively higher reported frequencies (>3.5%) in some studies5, 10, 16, 19, 22, 23 might have been due to loose criteria in population-based cancer registry,23 referral nature of hospital,5, 10, 16, 19, 22 or small case numbers with a chance variation5, 10, 16, 19 (Supplemental Table 1) Above all, synchronous cancer cases are rare, which has posed a difficulty in conducting a study with an adequate statistical power. Although one population-based cancer registry reported 596 synchronous colorectal cancer patients, tumoral molecular data were unavailable.23
Previous studies have examined a number of molecular events within synchronous colorectal cancer pairs.5–11 While some studies showed concordant patterns of MSI status within synchronous cancer pairs,8–10 others showed discordant patterns of MSI,7MLH1 methylation,10, 11 MSH2 expression,11 or p53 expression.5 Together with previous data,9–11 our results (i.e., a concordant CpG island methylation pattern in proximal/proximal and distal/distal pairs and a discordant methylation pattern in proximal/distal pairs) suggest both random and nonrandom components in CpG island methylation.
LINE-1 methylation level has been correlated with global DNA methylation level.46, 47 We have previously reported that LINE-1 hypomethylation is associated with poor prognosis among colon cancer patients,14 and inversely associated with CIMP-high and MSI-high.38 Global DNA methylation level may affect regulation of many proto-oncogenes and anti-oncogenes, and determine clinical, pathologic and molecular characteristics of colorectal cancer.14 In the current study, we demonstrated a significant correlation of LINE-1 methylation levels within synchronous cancer pairs, which might have been due to field effect. The concept of field effect (field defect or field cancerization) has been proposed to explain the development of multiple (synchronous or metachronous) primary tumors in the same organ.4 As a potential mechanism of field effect, epigenetic alterations have been proposed,4 and previous studies have reported the field effect due to MGMT promoter methylation or other CpG island methylation in normal appearing colonic mucosa.2 A study has revealed that LINE-1 methylation level is lower in some colon cancers and adjacent normal colon mucosa than peripheral blood from the same patients or normal colon mucosa in healthy individuals.48 However, in the current study, LINE-1 methylation level was consistently high in the adjacent normal mucosa that we tested, even in the cases with LINE-1 hypomethylation in both synchronous tumors. Likewise, neither CpG island methylation nor BRAF mutation was detected in the adjacent normal colon from synchronous cases. While these results do not support the field effect hypothesis, these do not exclude it either. Distribution of field effect may be patchy, or alternatively, there may exist an unidentified molecular abnormality in normal mucosa, which drives the development of the described alterations in LINE-1, CpG islands or BRAF. Additional studies are necessary to clarify the reasons why synchronous cancer pairs tend to show concordant patterns of some molecular alterations.
In our current study, compared to solitary cancers, synchronous cancers were more likely BRAF-mutated, CIMP-high and MSI-high. Accumulating evidence suggests that CIMP-high MSI-high colorectal cancer arise through the serrated pathway to colorectal cancer, which is characterized by a high frequency of BRAF mutation.49 Thus, compared to solitary cancers, synchronous colorectal cancers more likely arise through the serrated pathway, and possibly from multiple hyperplastic polyps or polyposis.3, 49 In addition, synchronous colorectal cancer pairs exhibit a significant correlation of LINE-1 methylation levels and concordant CpG island methylation, suggesting that shared environmental/genetic background may cause concordant patterns of DNA methylation and epigenetic events.
Studying clinical outcome and molecular changes is important in cancer research.50, 51 While most studies on clinical outcome of synchronous cancer patients17, 19–24 have reported no significant prognostic association with synchronous colorectal cancers, one study18 reported worse survival associated with synchronous cancers. However, in all of these studies,17–22, 24 controls were retrospectively selected; hence selection bias affecting clinical outcome data was a major concern. In contrast, our study represents the first prospective study to examine a survival difference between synchronous and solitary colorectal cancer patients. Our solitary cancers had arisen in the same background population as the synchronous cancer cases, thus eliminating sources of considerable bias that were inevitably present in any retrospective case-control studies.
Notably, one study23 utilized the population-based cancer registry, and found no significant survival difference between synchronous cases and solitary cancers. In that study,23 the proportion of synchronous cases was estimated to be 3.8% (596/15562), which was considerably higher than most previous studies5, 9, 17, 18, 20, 21, 24–26 and our current study (2.3%). It may be possible that synchronous cancer cases in the cancer registry study23 might have included patients with one cancer and a second malignant polyp [intramucosal carcinoma (pTim) or carcinoma in situ (pTis)]. We strictly limited synchronous cancer cases to those with two or more unequivocally separate cancers with at least submucosal invasion (pT1).
Possible reasons for the high mortality associated with synchronous cases remain speculative. There are several hypotheses (which are not necessarily mutually exclusive):
In summary, our prospective cohort study suggests that synchronous colorectal cancer cases are associated with poor prognosis. Compared to solitary cancers, synchronous colorectal cancers more commonly show BRAF mutations, CIMP-high and MSI-high. In addition, synchronous colorectal cancer pairs exhibit a significant correlation of LINE-1 methylation levels and concordant CpG island methylation. Our study may be an important one step forward to elucidate clinical and molecular characteristics of synchronous colorectal cancers.
Funding: This work was supported by the U.S. National Institute of Health (P01 CA87969 to S. Hankinson, P01 CA55075 to W. Willett, P50 CA127003 to C.S.F., K07 CA122826 to S.O.); the Bennett Family Fund; the Entertainment Industry Foundation National Colorectal Cancer Research Alliance. K.N. was supported by a fellowship grant from the Japan Society for Promotion of Science. The content is solely the responsibility of the authors and does not necessarily represent the official views of NCI or NIH. Funding agencies did not have any role in the design of the study; the collection, analysis, or interpretation of the data; the decision to submit the manuscript for publication; or the writing of the manuscript.
We deeply thank the Nurses’ Health Study and Health Professionals Follow-up Study cohort participants who have generously agreed to provide us with biological specimens and information through responses to questionnaires; hospitals and pathology departments throughout the U.S. for providing us with tumor tissue materials; Frank Speizer, Walter Willett, Susan Hankinson, Meir Stampfer, and many other staff members who implemented and have maintained the cohort studies.
Each author’s contribution: study concept and design (SO); acquisition of data (KN, SK, NI, KS, YB, JAM, ELG, CSF, SO); analysis and interpretation of data (KN, SK, NI, DS, ELG, CSF, SO); drafting of the manuscript (KN, SK, SO); critical revision of the manuscript for important intellectual content (KN, DS, JAM, ELG, CSF, SO); statistical analysis (KN, SK, DS, SO); funding support (CSF, SO); final approval of manuscript (all authors).
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.