Abstract
MicroRNAs (miRNAs) regulate up to one‐third of all protein‐coding genes including genes relevant to cancer. Variants within miRNAs have been reported to be associated with prognosis, survival, response to chemotherapy across cancer types, in vitro parameters of cell growth, and altered risks for development of cancer. Five miRNA variants have been reported to be associated with risk for development of colorectal cancer (CRC). In this study, we evaluated germline genetic variation in 1,123 miRNAs in 899 individuals with CRCs categorized by clinical subtypes and in 204 controls. The role of common miRNA variation in CRC was investigated using single variant and miRNA‐level association tests. Twenty‐nine miRNAs and 30 variants exhibited some marginal association with CRC in at least one subtype of CRC. Previously reported associations were not confirmed (n = 4) or could not be evaluated (n = 1). The variants noted for the CRCs with deficient mismatch repair showed little overlap with the variants noted for CRCs with proficient mismatch repair, consistent with our evolving understanding of the distinct biology underlying these two groups. © 2016 The Authors Genes, Chromosomes & Cancer Published by Wiley Periodicals, Inc.
INTRODUCTION
MicroRNAs (miRNAs) are a large group of noncoding RNAs, discovered in 1993 in Caenorhabditis elegans, now documented in many other organisms. They act in trans‐ upon cis‐regulatory elements in target messenger RNAs, affecting protein translation (Lee et al., 1993; Wightman et al., 1993). MiRNAs are located in introns of other genes and intergenic regions of the genome. Genes encoding miRNAs account for approximately 2% of the coding genes, and regulate up to 30% of all protein‐coding genes, notably regulating pathways relevant to cancer such as cell growth, differentiation, and apoptosis. MiRNAs can function as tumor‐suppressor genes or oncogenes, depending on whether deleted or overexpressed (Hayashita et al., 2005; Esquela‐Kerscher and Slack, 2006; Croce, 2009; Medina et al., 2010; Salzman and Weidhaas, 2013). Approximately 50% of miRNAs are in fragile regions of the genome that are often deleted, amplified, or misexpressed in cancers (Calin and Croce, 2006). MiRNAs are collated in a publicly available database, miRbase (http://www.mirbase.org/index.shtml). The most recent version (21; accessed May 12, 2015) contains 28,645 entries representing hairpin precursor miRNAs, expressing 35,828 mature miRNA products, in 223 different species. For Homo sapiens, 1,881 unique miRNAs are listed.
The sequence of an individual miRNA, on average 20–24 nucleotides long, determines its target, complementary to a portion of the 3′ UTR of the target gene's mRNA. Nucleotides 2–7 from the 5′ end of the miRNA are the major determinants of target selection for inhibition of expression. MiRNAs and target sites are highly conserved through evolution (Chen and Rajewsky, 2006) and single nucleotide polymorphisms (SNPs) located within miRNA are uncommon but do exist. Recent studies report associations between specific SNPs and prognosis, survival, response to chemotherapy across cancer types, in vitro parameters of cell growth, and risks for the development of cancer (Srivastava and Srivastava, 2012).
Exploration of an association between miRNA variants and colorectal cancer (CRC) has been limited. In this study, we estimated the frequency of germline SNPs and small insertions or deletions (indels) in miRNA using a targeted sequencing procedure. Association with CRC was evaluated for five categories of individuals diagnosed with CRC compared with controls. We also compared our findings with published literature on CRC associated with miRNA variants.
MATERIALS AND METHODS
Study Samples
Study samples were from the Colon Cancer Family Registry (Colon CFR), described in detail elsewhere (Newcomb et al., 2007) and at http://coloncfr.org. Between 1997 and 2012, the Colon CFR recruited families via both population‐based probands, recently diagnosed CRC cases from state or regional cancer registries in Australia, the USA, and Canada as well as clinic‐based probands enrolled from multiple‐case families referred to family‐cancer clinics in the same countries. Samples in this study were collected from the Australasian Colorectal Cancer Family Registry (Melbourne, Victoria, Australia), Hawaii Family Registry of Colon Cancer (Honolulu, HI), Mayo Colorectal Family Registry (Rochester, MN), Ontario Familial Colorectal Cancer Registry (Toronto, Ontario, Canada), Seattle Familial Colorectal Cancer Registry (Seattle, WA), and University of Southern California Consortium (Los Angeles, CA). Mismatch repair (MMR) status for all tumors was established, as previously described (Ait Ouakrim et al., 2015). All participants provided informed consent. Protocols were approved by the Institutional Review Board at each site.
Sequencing
MiRBase was used to identify 1,424 miRNAs for sequencing of the entire pre‐miRNA (http://www.mirbase.org/, build 17; see Supporting information).
Bioinformatics Analysis
Details of bioinformatics analysis are shown in the Supporting information.
Quality Control
Comprehensive quality control (QC) identified poor quality samples and potential sequencing batch effects. We investigated per‐sample percent duplicated reads, coverage of the capture region, variant calling quality and depth, variant call‐rate in the capture region, heterozygosity rate, transition:transversion ratio, and sex verification using PLINK/SEQ v0.10 (https://atgu.mgh.harvard.edu/plinkseq/index.shtml) and PLINK v1.9 (https://www.cog-genomics.org/plink2). Sample contamination was visually inspected by plotting the fraction of ALT reads at common variant positions against the 1000 Genomes Project allele frequency; samples displaying more than three bands or a “shotgun pattern” indicate probable contamination or poor quality DNA. Pedigree Relationship Statistical Test‐Plus was used to identify related samples and population stratification was evaluated using STRUCTURE software (Patterson et al., 2006; Price et al., 2006). Samples with <90% of the capture region covered at 10X, call rate within the capture region <95%, suspected sample contamination, unexpected familial relationships, and <80% European ancestry were excluded from analysis. For variant quality filtering, we included GATK VQSR filtering tranche 99.0 and above. Polymorphic variants mapping to five or more locations in the genome, those with call rate <95%, monomorphic, and those with Hardy–Weinberg equilibrium P value <1E−5 in our controls were excluded.
Analysis
Six case–control analyses were performed, comparing five CRC case sets and all cases combined to the combined group of controls. For CRC cases for whom tumor testing had been conducted, each was categorized as having deficient or proficient DNA mismatch repair tumors (dMMR and pMMR, respectively) (Lindor et al., 2002). We defined five categories of cases: those with dMMR tumors for which no germline mutation could be identified (dMMR); familial colorectal cancer type X cases (Lindor et al., 2006) combined with those from other pMMR multi‐case‐CRC families from a prior linkage study, not otherwise specified (FCCTX/pMMR linkage); pMMR CRC diagnosed before age 50 years (Young Onset); those for which no tumor had been available for MMR characterization and no causal MMR gene mutation had been found by sequencing (“unselected” [referring to tumor MMR status which was unknown as no tumor was available for testing]); and a combined group (Likely pMMR) that included nonoverlapping cases from the FCCTX/pMMR linkage, Young Onset, and unselected cases (Table 1). To increase power, “controls”: included non‐carrier spouses and MMR carriers, the cause of whose CRC is considered known. We used principal components analysis as implemented in the SNPRelate R package (Zheng et al., 2012) to evaluate sample eigenvectors as covariates to adjust for possible population stratification. The SNPs used for principal components analysis included approximately 2,000 common (MAF >5%), independent (linkage disequilibrium r 2 < 0.4), autosomal SNPs. None of the top eigenvectors was associated with case‐control status indicating that no population stratification adjustment was necessary.
Table 1.
Numbers of Individuals Within Different Categories Used in Analyses
| Group | Original, n | Passed quality control, n | Quality control and Europeana, n |
|---|---|---|---|
| Controls | 204 | ||
| Mismatch repair carrier control | 165 | 163 | 113 |
| Noncarrier spousal control | 95 | 91 | 91 |
| dMMR, no mutation | 147 | 147 | 129 |
| FCCTX/pMMR linkage | 288 | 285 | 229 |
| Young Onset | 234 | 234 | 206 |
| Unselected | 602 | 602 | 335 |
| Likely pMMRb | 1,076 | 1,070 | 734 |
| Total assigned cases | 1,271 | 1,265 | 899 |
European subset defined as samples with >80% European ancestry based on STRUCTURE.
Combination of FCCTX/pMMR linkage, Young Onset, and unselected cases.
We performed both single‐SNP‐ and miRNA‐level analyses using an extension to commonly used gene‐based statistics to allow for known pedigree relationships (Schaid et al., 2013). For miRNA‐level tests, analyses were conducted using both a burden test (most powerful if variants in a gene have effects in the same direction) and kernel statistic (most powerful if variants have effects in opposite directions). Variants were weighted using beta density weights of (1, 25), with rare variants receiving a higher weight. False discovery rate was calculated using the R package Q‐value (Storey et al., 2015) and considering all case–control comparisons.
Our approach allows for both pedigree data, for example, multiple cases from a family as well as unrelated subjects and takes a retrospective view treating the trait as fixed and genotypes as random, allowing complex and undefined ascertainment of pedigrees as is typical for many of the pedigrees included in our study.
We conducted a literature search for miRNA variants reported to be associated with altered risk for CRC. The frequency of these variants was determined in all our subtypes and controls.
RESULTS
A total of 1,436 individuals with CRC and 95 unaffected spouse controls were selected (Table 1). Those with CRC included 165 CRCs in individuals with known MMR germline mutations (which were used as mutation‐positive controls). After removing samples failing QC, unexpected duplicate results, cryptically related individuals, and non‐European ancestry, 1,103 subjects were included in the analysis. Final comparison groups were All cases (n = 899), with case subsets including dMMR (n = 129), FCCTX/pMMR Linkage (n = 229), Young Onset (n = 206), unselected (n = 335), and Likely pMMR (n = 734) cases. Spouses and affected MMR carriers served as controls (n = 204) for case–control comparisons.
A total of 1,316 variants in 689 miRNA passed QC filters and variants in 575 miRNAs were polymorphic in European samples and were included in analysis. Three hundred eighty miRNA had more than a single variant available for analysis and were included in miRNA‐level analyses, while 242 variants in 195 miRNA with MAF >1% were analyzed for single‐variant association. The average number of variants per miRNA included in miRNA‐level analysis was 3.1 (range 2–24) with over half (n = 210, 55%) of miRNA including only two variants. Considering multiple testing, none of the miRNA‐level tests was statistically significant (minimum miRNA‐level P value = 0.003, false discovery rate = 0.41). MiRNA exhibiting a marginal positive association in at least one CRC subtype are presented in Table 2 where “marginal association” is defined as a higher frequency of rare variants in cases (positive‐burden statistic) and a kernel statistic P value <0.10. Single variants meeting these same criteria are presented in Table 3. The miRNA location of the variant and the frequency of that variant in public databases is included in Table S1. The frequency of variants with minor allele frequency >1% in our controls matched that of the 1000 Genomes Project well (Fig. S1).
Table 2.
| miRNA | Chromosome | Start | Stop | All cases (n = 899) | dMMR cases (n = 129) | FCCTX/pMMR linkage cases (n = 229) | Young Onset cases (n = 206) | Unselected cases (n = 335) | pMMR cases (n = 734) | Minimum FDR Q‐valued | ||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Number of variants | Signed kernel P valuec | Number of variants | Signed kernel P valuec | Number of variants | Signed kernel P valuec | Number of variants | Signed kernel P valuec | Number of variants | Signed kernel P valuec | Number of variants | Signed kernel P valuec | |||||
| MIR1262 | chr1 | 68649200 | 68649293 | 2 | 0.0802 | – | – | – | – | – | – | 2 | 0.0421 | 2 | 0.0735 | 0.410 |
| MIR216A | chr2 | 56216084 | 56216194 | 2 | 0.1108 | – | – | 2 | 0.3373 | – | – | – | – | 2 | 0.0985 | 0.410 |
| MIR3679 | chr2 | 134884695 | 134884763 | 8 | 0.1861 | 5 | 0.0255 | 7 | 0.1898 | 5 | 0.1346 | 6 | 0.6089 | 8 | 0.2638 | 0.410 |
| MIR3138 | chr4 | 10080234 | 10080316 | 4 | 0.0640 | 4 | 0.3276 | 4 | −0.3837 | 3 | 0.1170 | 3 | 0.0258 | 4 | 0.0965 | 0.410 |
| MIR1289‐2 | chr5 | 132763287 | 132763398 | 3 | −0.5971 | 3 | 0.0049 | 3 | 0.7297 | 2 | −0.6346 | 2 | −0.0254 | 3 | −0.3347 | 0.410 |
| hsa‐mir‐1294 | chr5 | 153726665 | 153726807 | 4 | 0.2487 | 2 | 0.2606 | 4 | 0.7498 | 2 | 0.0913 | 2 | 0.4335 | 4 | 0.2404 | 0.410 |
| MIR4465 | chr6 | 141004950 | 141005020 | 6 | 0.4849 | 6 | 0.0766 | – | – | – | – | – | – | – | – | 0.410 |
| MIR4470 | chr8 | 62627346 | 62627418 | 5 | 0.3635 | 2 | 0.0748 | 2 | 0.4219 | 4 | 0.4114 | 2 | 0.2112 | 4 | 0.3937 | 0.410 |
| MIR4289 | chr9 | 91360750 | 91360820 | 5 | 0.1443 | 4 | 0.8337 | 4 | 0.0462 | 3 | 0.1564 | 4 | 0.1093 | 5 | 0.0887 | 0.410 |
| MIR199B | chr9 | 131006999 | 131007109 | 3 | 0.1228 | 2 | 0.0648 | 2 | 0.1440 | 2 | −0.3654 | 3 | 0.0699 | 3 | 0.1675 | 0.410 |
| MIR129‐2 | chr11 | 43602943 | 43603033 | 2 | 0.0606 | – | – | – | – | – | – | 2 | 0.1349 | – | – | 0.410 |
| MIR612 | chr11 | 65211928 | 65212028 | 2 | 0.9413 | 2 | 0.7416 | 2 | −0.1764 | 2 | 0.0432 | 2 | −0.5867 | 2 | −0.9554 | 0.410 |
| MIR381HG | chr14 | 101511493 | 101518132 | 14 | 0.0684 | 7 | 0.2151 | 10 | 0.1382 | 9 | 0.0058 | 9 | 0.3512 | 14 | 0.0826 | 0.410 |
| MIR656 | chr14 | 101533060 | 101533138 | 3 | 0.1277 | 2 | 0.0852 | – | – | 2 | 0.1137 | – | – | 2 | 0.1260 | 0.410 |
| MIR1225 | chr16 | 2140195 | 2140285 | 3 | 0.4868 | 2 | 0.0518 | 2 | −0.7932 | 2 | −0.8113 | 3 | −0.7588 | 3 | −0.6683 | 0.410 |
| MIR4520A | chr17 | 6558758 | 6558828 | 3 | 0.1125 | 3 | 0.2301 | 2 | 0.2811 | 2 | −0.7596 | 3 | 0.0132 | 3 | 0.1349 | 0.410 |
| MIR4520B | chr17 | 6558767 | 6558821 | 3 | 0.1125 | 3 | 0.2301 | 2 | 0.2811 | 2 | −0.7596 | 3 | 0.0132 | 3 | 0.1349 | 0.410 |
| MIR4743 | chr18 | 46196970 | 46197039 | 2 | 0.2919 | 2 | 0.0477 | 2 | 0.4868 | 2 | 0.4817 | 2 | 0.1682 | 2 | 0.4333 | 0.410 |
| MIR3190 | chr19 | 47730198 | 47730278 | 2 | 0.2308 | – | – | – | – | – | – | 2 | 0.0107 | 2 | 0.1658 | 0.410 |
| MIR3192 | chr20 | 18451258 | 18451335 | 2 | 0.0385 | – | – | – | – | – | – | 2 | 0.0094 | 2 | 0.0387 | 0.410 |
| MIR499A | chr20 | 33578178 | 33578300 | 6 | 0.2572 | 2 | 0.0770 | 3 | 0.4153 | 2 | 0.0476 | 5 | 0.6552 | 6 | 0.3162 | 0.410 |
| MIR941‐3 | chr20 | 62550833 | 62550947 | 4 | 0.0884 | – | – | 3 | 0.0731 | – | – | 2 | 0.1349 | 4 | 0.1187 | 0.410 |
| MIR941‐4 | chr20 | 62550836 | 62550950 | 5 | 0.0981 | 4 | 0.0283 | 4 | 0.1720 | 4 | 0.1425 | 5 | 0.4872 | 5 | 0.2206 | 0.410 |
| MIR941‐3 | chr20 | 62550889 | 62551003 | 4 | 0.0884 | – | – | 3 | 0.0731 | – | – | 2 | 0.1349 | 4 | 0.1187 | 0.410 |
| MIR941‐4 | chr20 | 62550892 | 62551006 | 5 | 0.0981 | 4 | 0.0283 | 4 | 0.1720 | 4 | 0.1425 | 5 | 0.4872 | 5 | 0.2206 | 0.410 |
| MIR941‐3 | chr20 | 62551084 | 62551201 | 4 | 0.0884 | – | – | 3 | 0.0731 | – | – | 2 | 0.1349 | 4 | 0.1187 | 0.410 |
| MIR941‐4 | chr20 | 62551196 | 62551313 | 5 | 0.0981 | 4 | 0.0283 | 4 | 0.1720 | 4 | 0.1425 | 5 | 0.4872 | 5 | 0.2206 | 0.410 |
| MIR3687 | chr21 | 9826202 | 9826263 | 5 | 0.1122 | 3 | 0.1398 | 3 | 0.2864 | 5 | 0.3865 | 3 | 0.0980 | 5 | 0.1230 | 0.410 |
| MIR658 | chr22 | 38240278 | 38240378 | 3 | 0.7210 | 2 | 0.7436 | 2 | 0.9663 | 3 | 0.0778 | 2 | −0.8359 | 3 | 0.5887 | 0.410 |
Marginal evidence defined as sign of the burden statistic multiplied by kernel statistic P value in [0, 0.10].
The control group consists of 91 spousal controls and 113 participants with known mismatch repair gene mutations (n = 204 total).
Highlighted (bold) boxes = p value <0.10.
FDR, false discovery rate.
Table 3.
| Signed kernel P value | Minimum FDR Q‐valueg | ||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|
| miRNA | Chromosome | POSc | REFd | ALTe | All cases (n = 899)f | dMMR cases (n = 129)f | FCCTX/pMMR linkage cases (n = 229)f | Young Onset cases (n = 206)f | Unselected cases (n = 335)f | pMMR cases (n = 734)f | |
| MIR216A | chr2 | 56216090 | A | T | 0.102 | 0.142 | 0.320 | 0.138 | 0.117 | 0.090 | 0.271 |
| MIR663B | chr2 | 133014587 | C | T | 0.488 | 0.016 | −0.630 | −0.665 | 0.331 | 0.955 | 0.271 |
| MIR1258 | chr2 | 180725568 | T | C | 0.519 | 0.027 | 0.821 | −0.780 | 0.759 | 0.787 | 0.271 |
| MIR4268 | chr2 | 220771223 | C | T | 0.222 | 0.407 | 0.064 | 0.745 | 0.403 | 0.230 | 0.271 |
| MIR4789 | chr3 | 175087408 | C | T | 0.099 | −0.525 | 0.016 | 0.091 | 0.224 | 0.046 | 0.271 |
| hsa‐mir‐1294 | chr5 | 153726769 | A | G | 0.265 | 0.241 | 0.789 | 0.088 | 0.459 | 0.250 | 0.271 |
| hsa‐mir‐3144 | chr6 | 120336327 | C | A | 0.166 | 0.643 | 0.033 | 0.852 | 0.191 | 0.123 | 0.271 |
| MIR4467 | chr7 | 102111936 | G | A | 0.062 | 0.289 | 0.034 | 0.076 | 0.289 | 0.062 | 0.271 |
| MIR3622A:MIR3622B | chr8 | 27559214 | G | A | 0.104 | 0.066 | 0.278 | 0.117 | 0.271 | 0.138 | 0.271 |
| hsa‐mir‐1302‐7 | chr8 | 142867668 | ATGT | A | 0.022 | 0.012 | 0.107 | 0.026 | 0.009 | 0.021 | 0.271 |
| MIR4669 | chr9 | 137271318 | C | A | 0.008 | 0.173 | 0.109 | 0.003 | 0.030 | 0.007 | 0.271 |
| MIR3689A | chr9 | 137742206 | C | T | 0.050 | 0.354 | 0.049 | 0.248 | 0.038 | 0.043 | 0.271 |
| MIR1908 | chr11 | 61582708 | T | C | 0.841 | 0.016 | −0.636 | −0.424 | 0.885 | −0.713 | 0.271 |
| MIR612 | chr11 | 65211940 | C | A | 0.930 | 0.745 | −0.174 | 0.043 | −0.590 | −0.942 | 0.271 |
| MIR492 | chr12 | 95228286 | G | C | 0.199 | 0.078 | 0.077 | 0.905 | 0.435 | 0.284 | 0.271 |
| hsa‐mir‐300:MIR300 | chr14 | 101507727 | C | T | 0.235 | 0.778 | 0.213 | 0.074 | 0.619 | 0.198 | 0.271 |
| MIR381HG | chr14 | 101513795 | C | T | 0.042 | 0.138 | 0.095 | 0.003 | 0.305 | 0.048 | 0.271 |
| MIR656 | chr14 | 101533093 | C | T | 0.118 | 0.087 | 0.115 | 0.110 | 0.369 | 0.119 | 0.271 |
| MIR4513 | chr15 | 75081078 | G | A | 0.189 | 0.087 | 0.314 | −0.798 | 0.095 | 0.324 | 0.271 |
| MIR184 | chr15 | 79502168 | G | T | 0.251 | 0.511 | 0.479 | 0.749 | 0.097 | 0.336 | 0.271 |
| MIR1225 | chr16 | 2140262 | T | TC | 0.567 | 0.047 | 0.913 | −0.988 | 0.868 | 0.831 | 0.271 |
| MIR4520A:MIR4520B | chr17 | 6558808 | G | A | 0.104 | 0.241 | 0.279 | −0.766 | 0.012 | 0.127 | 0.271 |
| MIR423 | chr17 | 28444183 | A | C | 0.059 | 0.168 | 0.833 | 0.203 | 0.006 | 0.088 | 0.271 |
| MIR4745 | chr19 | 804959 | C | T | 0.718 | −0.759 | −0.614 | 0.079 | 0.941 | 0.605 | 0.271 |
| MIR3190:MIR3191 | chr19 | 47730272 | A | C | 0.186 | 0.811 | −0.398 | −0.043 | 0.00008 | 0.126 | 0.155 |
| MIR4751 | chr19 | 50436371 | G | A | 0.121 | 0.092 | 0.148 | 0.382 | 0.243 | 0.139 | 0.271 |
| MIR4754 | chr19 | 58898193 | C | T | 0.001 | 0.164 | 0.003 | 0.018 | 0.00100 | 0.0005 | 0.271 |
| MIR3192 | chr20 | 18451325 | T | C | 0.036 | 0.135 | 0.085 | 0.182 | 0.009 | 0.036 | 0.271 |
| hsa‐mir‐941‐3:MIR941‐2:MIR941‐4 | chr20 | 62551298 | C | T | 0.264 | 0.071 | 0.567 | 0.174 | 0.672 | 0.445 | 0.271 |
| MIRLET7BHG | chr22 | 46487011 | G | A | 0.104 | 0.034 | 0.159 | 0.259 | 0.313 | 0.123 | 0.271 |
Marginal evidence defined as sign of the burden statistic multiplied by kernel statistic P value in [0, 0.10].
The control group consists of 91 spousal controls and 113 participants with known mismatch repair gene mutations.
POS, genomic position.
REF, reference allele.
ALT, alternate allele.
Highlighted (bold) boxes = P value <0.10.
FDR, false discovery rate. Minimum FDR Q‐value for each variant. FDR Q‐value was calculated considering all case–control comparisons simultaneously.
Analysis of the same CRC subtypes using only the spousal controls was conducted but did not substantively change results (results not shown). The decision to combine the MMR positive DNAs with the spouse controls was based on the reasonable hypothesis that CRC in individuals with known MMR deficiency was explained by the germline mutation and the probability of other contributing factors approximates that of the general population.
Five miRNA variants were found in the literature reporting altered risks for CRC including rs11614913 (miRNA196a2), rs2910164 (miRNA146a), rs4938723 (miRNA34b/c), rs2292832 (miRNA149), and rs3746444 (miRNA499; Table S2). Our results did not confirm these associations in the four variants we could evaluate. The fifth variant, rs4938723 in miRNA34b/c, had no coverage in our dataset, perhaps indicating a failure in the sequencing capture.
DISCUSSION
In this observational case‐control study, we sought to evaluate whether variants that occur in miRNA genes were associated with CRC. Of the 1,424 different miRNAs studied, 29 miRNAs and 30 variants exhibited some marginal association in at least one subtype of CRC. No variant was associated with all subtypes of CRC (Tables 2 and 3). The miRNAs of interest (albeit marginally significant) were not found in previous studies. It is notable that our subgroup with definite dMMR tumors exhibited association with miRNAs that had little overlap with the miRNAs associated with the other predominantly pMMR groups. This is not unexpected based upon knowledge of the fundamentally different underlying biology of the dMMR group, recently reaffirmed by new definitions of consensus molecular subtypes (Guinney et al., 2015). We acknowledge that we had limited power to detect overlap. Therefore, although lack of overlap may be consistent with different biology, it does not confirm it. The results of the present study do support the importance of conducting research that does not ignore the well‐defined molecular heterogeneity of CRC.
Five miRNA variants have been associated with altered risk for CRC in some but not all studies (Table S3). We looked specifically at these five variants: one was not well captured in our dataset so could not be evaluated, but the other four were not different between cases and controls. It is notable that the majority of the published studies to date were conducted in Asian populations whereas our study was restricted to Europeans; it is possible these variants are in linkage disequilibrium with an ethnic‐specific risk factor or that our study was underpowered to detect a modest association. Other investigators also report non‐replication of these miRNAs in CRC cohorts of European ancestry (Hezova et al., 2012; Vinci et al., 2013; Kupcinskas et al., 2014). In addition, expression levels for these miRNAs were not reported to differ in CRC across the newly described consensus molecular subtypes of CRC (Guinney et al., 2015). Larger studies with careful attention to ethnic selection are needed to assess the validity of all observations.
One strength of this study was the ability to evaluate across well‐characterized subsets of CRC cases (dMMR, pMMR, Young Onset, etc.) for whom other major germline mutations had been sought but were not found. Second, coverage of miRNAs was broad due to the inclusion of nearly all the miRNAs known at the time the study was initiated Overall, the quality of the sequencing reads was high and 1,123 miRNAs could be evaluated. One weakness was our limited sample size and the number of controls. However, our control allele frequencies matched frequencies in the 1000 Genomes European Project well (Fig. S1). Another weakness is the absence of functional studies to follow‐up on our current findings, which is beyond the scope of this short report.
This study identified a list of miRNAs for which there was a suggestion of association with CRC, which varied by molecular subtype. These findings argue for additional testing in a larger study. We have provided an assessment using a European sample of four of the miRNA variants reported by others to be associated with CRC and were not able to confirm those associations even though our numbers were comparable to the discovery reports.
Supporting information
Supporting Information
REFERENCES
- Ait Ouakrim D, Dashti SG, Chau R, Buchanan DD, Clendenning M, Rosty C, Winship IM, Young JP, Giles GG, Leggett B, Macrae FA, Ahnen DJ, Casey G, Gallinger S, Haile RW, Le Marchand L, Thibodeau SN, Lindor NM, Newcomb PA, Potter JD, Baron JA, Hopper JL, Jenkins MA, Win AK. 2015. Aspirin, ibuprofen, and the risk for colorectal cancer in Lynch Syndrome. J Natl Cancer Inst 107. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Calin GA, Croce CM. 2006. MicroRNA signatures in human cancers. Nat Rev Cancer 6:857–866. [DOI] [PubMed] [Google Scholar]
- Chen K, Rajewsky N. 2006. Deep conservation of microRNA‐target relationships and 3′UTR motifs in vertebrates, flies, and nematodes. Cold Spring Harbor Symp Quant Biol 71:149–156. [DOI] [PubMed] [Google Scholar]
- Croce CM. 2009. Causes and consequences of microRNA dysregulation in cancer. Nat Rev Genet 10:704–714. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Esquela‐Kerscher A, Slack FJ. 2006. Oncomirs—microRNAs with a role in cancer. Nat Rev Cancer 6:259–269. [DOI] [PubMed] [Google Scholar]
- Guinney J, Dienstmann R, Wang X, de Reynies A, Schlicker A, Soneson C, Marisa L, Roepman P, Nyamundanda G, Angelino P, Bot BM, Morris JS, Simon IM, Gerster S, Fessler E, De Sousa EMF, Missiaglia E, Ramay H, Barras D, Homicsko K, Maru D, Manyam GC, Broom B, Boige V, Perez‐Villamil B, Laderas T, Salazar R, Gray JW, Hanahan D, Tabernero J, Bernards R, Friend SH, Laurent‐Puig P, Medema JP, Sadanandam A, Wessels L, Delorenzi M, Kopetz S, Vermeulen L, Tejpar S. 2015. The consensus molecular subtypes of colorectal cancer. Nat Med 21:1350–1356. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Hayashita Y, Osada H, Tatematsu Y, Yamada H, Yanagisawa K, Tomida S, Yatabe Y, Kawahara K, Sekido Y, Takahashi T. 2005. A polycistronic microRNA cluster, miR‐17‐92, is overexpressed in human lung cancers and enhances cell proliferation. Cancer Res 65:9628–9632. [DOI] [PubMed] [Google Scholar]
- Hezova R, Kovarikova A, Bienertova‐Vasku J, Sachlova M, Redova M, Vasku A, Svoboda M, Radova L, Kiss I, Vyzula R, Slaby O. 2012. Evaluation of SNPs in miR‐196‐a2, miR‐27a and miR‐146a as risk factors of colorectal cancer. World J Gastroenterol 18:2827–2831. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Kupcinskas J, Bruzaite I, Juzenas S, Gyvyte U, Jonaitis L, Kiudelis G, Skieceviciene J, Leja M, Pauzas H, Tamelis A, Pavalkis D, Kupcinskas L. 2014. Lack of association between miR‐27a, miR‐146a, miR‐196a‐2, miR‐492 and miR‐608 gene polymorphisms and colorectal cancer. Sci Rep 4:5993. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Lee RC, Feinbaum RL, Ambros V. 1993. The C. elegans heterochronic gene lin‐4 encodes small RNAs with antisense complementarity to lin‐14. Cell 75:843–854. [DOI] [PubMed] [Google Scholar]
- Lindor NM, Burgart LJ, Leontovich O, Goldberg RM, Cunningham JM, Sargent DJ, Walsh‐Vockley C, Petersen GM, Walsh MD, Leggett BA, Young JP, Barker MA, Jass JR, Hopper J, Gallinger S, Bapat B, Redston M, Thibodeau SN. 2002. Immunohistochemistry versus microsatellite instability testing in phenotyping colorectal tumors. J Clin Oncol 20:1043–1048. [DOI] [PubMed] [Google Scholar]
- Lindor NM, Petersen GM, Hadley DW, Kinney AY, Miesfeldt S, Lu KH, Lynch P, Burke W, Press N. 2006. Recommendations for the care of individuals with an inherited predisposition to Lynch Syndrome: A systematic review. J Am Med Assoc 296:1507–1517. [DOI] [PubMed] [Google Scholar]
- Medina PP, Nolde M, Slack FJ. 2010. OncomiR addiction in an in vivo model of microRNA‐21‐induced pre‐B‐cell lymphoma. Nature 467:86–90. [DOI] [PubMed] [Google Scholar]
- Newcomb PA, Baron J, Cotterchio M, Gallinger S, Grove J, Haile R, Hall D, Hopper JL, Jass J, Le Marchand L, Limburg P, Lindor N, Potter JD, Templeton AS, Thibodeau S, Seminara D. 2007. Colon Cancer Family Registry: An international resource for studies of the genetic epidemiology of colon cancer. Cancer Epidemiol Biomarkers Prev 16:2331–2343. [DOI] [PubMed] [Google Scholar]
- Patterson N, Price AL, Reich D. 2006. Population structure and eigenanalysis. PLoS Genet 2:e190. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Price AL, Patterson NJ, Plenge RM, Weinblatt ME, Shadick NA, Reich D. 2006. Principal components analysis corrects for stratification in genome‐wide association studies. Nat Genet 38:904–909. [DOI] [PubMed] [Google Scholar]
- Salzman DW, Weidhaas JB. 2013. SNPing cancer in the bud: microRNA and microRNA‐target site polymorphisms as diagnostic and prognostic biomarkers in cancer. Pharmacol Therap 137:55–63. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Schaid DJ, McDonnell SK, Sinnwell JP, Thibodeau SN. 2013. Multiple genetic variant association testing by collapsing and kernel methods with pedigree or population structured data. Genet Epidemiol 37:409–418. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Srivastava K, Srivastava A. 2012. Comprehensive review of genetic association studies and meta‐analyses on miRNA polymorphisms and cancer risk. PLoS One 7:e50966. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Storey JD, Bass A, Dabney A, Robinson D. 2015. qvalue: Q‐Value Estimation for False Discovery Rate Control. R Package, Version 2.3.0 ed.
- Vinci S, Gelmini S, Mancini I, Malentacchi F, Pazzagli M, Beltrami C, Pinzani P, Orlando C. 2013. Genetic and epigenetic factors in regulation of microRNA in colorectal cancers. Methods 59:138–146. [DOI] [PubMed] [Google Scholar]
- Wightman B, Ha I, Ruvkun G. 1993. Posttranscriptional regulation of the heterochronic gene lin‐14 by lin‐4 mediates temporal pattern formation in C. elegans . Cell 75:855–862. [DOI] [PubMed] [Google Scholar]
- Zheng X, Levine D, Shen J, Gogarten SM, Laurie C, Weir BS. 2012. A high‐performance computing toolset for relatedness and principal component analysis of SNP data. Bioinformatics 28:3326–3328. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Supporting Information