Association between hsa-mir-146a genotype and tumor age-of-onset in BRCA1/BRCA2 -negative familial breast and ovarian cancer patients

Abstract

An increasing body of evidence points to a possible role of microRNAs (miRNAs) in hereditary cancer syndromes. To evaluate the role of miRNA allelic variants in the susceptibility to familial breast and ovarian cancers in BRCA1/BRCA2 -negative patients, we focused our attention on three miRNAs, miR-146a, miR-17 and miR-369, based on their affinity to either BRCA1 or BRCA2 messenger RNA and their localization on chromosome regions commonly deleted in those tumors. The analysis was performed on 101 Italian probands with ascertained familiarity for breast/ovarian cancer and tested negative for both BRCA1 and BRCA2 gene mutations.

No allelic variant was detected for hsa-mir-17 and hsa-mir-369 , and allelic and genotype frequencies for miR-146a rs2910164 single-nucleotide polymorphism (SNP) were comparable with that of 155 controls from the same population, ruling out a role for genetic variations in these three miRNAs as major determinants in cancer predisposition of BRCA1/BRCA2 -negative patients.

Instead, our study suggests that mir-146a rs2910164 SNP may impact on the age of cancer onset. In fact, subjects with mir-146a a GC or CC genotypes developed tumors at younger age compared with individuals with the GG genotype Thus, in contrast to a recent report, our data support the hypothesis by Shen and coworkers of an association between the C allele of hsa-mir-146a and early cancer onset and prompt further investigations on the relevance of this polymorphism in early familial breast/ovarian tumor development.

Introduction

About 5–10% of all breast cancers are estimated to arise from hereditary predisposition ( 1 , 2 ). Recent large-scale mutation analyses indicate that, in most populations, only ∼30 to 60% of familial breast cancers are attributable to mutations in either BRCA1 or BRCA2 , the two currently known major breast cancer predisposing genes. Therefore, in a substantial fraction of patients, other, still unidentified, factors are probably responsible for the increased familial risk ( 3 ).

According to the current polygenic model of inherited tumors, unfavorable combinations of polymorphic genetic variants in low-penetrance susceptibility genes are postulated to account for the excess familial cancer risk ( 4 , 5 ). Among these genes, recent evidence point at pre-microRNA (pre-miRNA) genes as possible candidates. In fact, it has been shown that miRNAs play an important role in cancer development by modulating the expression of both oncogenes and tumor suppressor genes ( 6 ). Based on this information, it has been suggested that individual variations in pre-miRNA genes that affect either the expression or the affinity of the mature miRNA to the target sequence may account for certain forms of familial cancer.

In accord with this hypothesis, Yang et al. ( 7 ) described a genetic variant in mir-27a that seems to associate with a reduced breast cancer risk in BRCA1/BRCA2 -negative patients. Shen et al. ( 8 , 9 ) recently identified rare variants in mir-30c-1 and mir-17 in BRCA1/BRCA2 -negative breast cancer patients and a novel variant in mir-191 in an ovarian cancer family, but the impact of these variations in cancer risk was undetermined. These same authors reported a correlation between a polymorphism in mir-146a and early cancer onset in familial BRCA1/BRCA2 -negative breast/ovarian cancer patients ( 10 ), but this association was subsequently negated by Catucci et al. ( 11 ), who also ruled out a role for miR-146a, miR-196a2 or miR-499 in cancer risk.

Given these conflicting data, we sought to investigate the role of three miRNAs, miR-146a, miR-17 and miR-369, on tumor predisposition in BRCA1/BRCA2 -negative patients. These miRNAs were selected based on their documented ability to bind the 3′-untranslated regions of either BRCA1 or BRCA2 ( hsa-mir-17 ; hsa-mir-146a ) ( 8 ) or predicted in silico to bind BRCA2 ( hsa-mir-369 ) ( http://mirdb.org/miRDB/index.html ; http://www.targetscan.org/ both last accessed 17/09/2010). Moreover, these miRNAs map to chromosome regions (13q31, 5q34 and 14q32, respectively) commonly deleted in both sporadic and familial breast and ovarian cancers ( 12 , 13 ).

Material and methods

From 2002 to 2008, a total of 348 index cases were analyzed for BRCA1 and BRCA2 mutation at the Centro di Riferimento Oncologico, CRO National Cancer Institute. All subjects were Italians. Eligibility criteria were essentially as in Santarosa et al. ( 14 ). In detail, probands were considered eligible for BRCA1/BRCA2 testing when fulfilling at least one the following minimum criteria: (i) three or more cases of breast and/or ovarian cancer at any age, with one case being a first-degree relative of the other two; (ii) two first-degree relatives with breast cancer diagnosed before 50 years of age or at any age but with one case of bilateral breast cancer; (iii) two first-degree relatives with ovarian cancer at any age or one ovarian cancer at any age and one breast cancer before the age of 50; (iv) cases of breast cancer in male; (v) breast and ovarian cancer in the same woman and (vi) breast cancer in women <30years. Second-degree relatives were considered when paternal transmission was evident. Written informed consent for genetic testing was obtained from all the participants. The study fully conformed to the Declaration of Helsinki and to the current Italian Legislation and was also approved by the Ethical Committee of the Institute.

Genomic DNA was purified from blood samples according to standard procedures. Screening for mutations in the BRCA1 and BRCA2 genes was carried out by denaturing high performance liquid chromatography (dHPLC) and polymerase chain reaction (PCR)-direct sequencing. Primer sequences and PCR conditions were basically as described previously ( 14 , 15 ), with slight modifications (available on request). Germ line rearrangements of BRCA1 and BRCA2 genes were also analyzed by multiple ligation-dependent probe amplification using SALSA MLPA Kits (MRC Holland, Amsterdam, The Netherlands).

PCR-direct sequencing was also used to identify mir-146a , mir-17 and mir-369 sequence variations, using the following primers and PCR conditions: miR17F 5′-CTAATTATCTATTTCAAATTTAGCA-3′/miR17R 5′-TGCACTTTAAAGCCCAACTT-3′, miR369F 5′-GTCCACTAGCCGTCCGTATC-3′/miR369R 5′-GGGTCCTCCAAGGTGAGATT-3′, miR146aF 5′-CCGATGTGTATCCTCAGCTTTG-3′/miR146aR 5′-GCCTGAGACTCTGCCTTCTG-3′, 30 cycles at 95°C for 30 s, 56°C ( mir-17 ) or 60°C ( mir-146a and mir-369 ) for 30 s and 72°C for 60 s using GoTaq polymerase and buffer by Promega (Madison, Wisconsin, USA). Analyses were performed blind of the patient’s characteristics and confirmed on both DNA strands.

Fisher’s exact test was used to compare genotype and allele frequencies between case and control groups. Deviation from Hardy–Weinberg equilibrium was tested by using the χ ² test for goodness of fit. The non-parametric Wilcoxon test was used to compare the age of cancer onset between genotype groups. The age at the first cancer diagnosis was considered in the case of tumor multiplicity. The association between genotype and age at disease onset was assessed also by the use of the Kaplan–Meier method. Homogeneity of the curves was evaluated by using log-rank and Wilcoxon tests.

Results and discussion

Of the 348 index cases tested for BRCA1 and BRCA2 mutations at the Centro di Riferimento Oncologico, CRO National Cancer Institute from 2002 to 2008, 71 (20.4%) carried mutations at either gene. Of the 277 probands negative for mutations in BRCA1/BRCA2 , 101 (95 females and 6 males) were selected for this study based on the availability of DNA and a high likelihood to be deficient for BRCA1/BRCA2 pathways according to a high BRCAPro score (>10%) ( http://astor.som.jhmi.edu/BayesMendel/brcapro.html last accessed 17/09/2010). Ninety-four cases were diagnosed with breast cancer (multiple breast tumors in 14 cases), two with both breast and ovarian cancer and five with ovarian cancer. As for family history, in the families of the 101 index cases, only diagnosis of breast cancers were reported in 81 cases, breast and ovarian cancer in 18 and only ovarian cancer in 2. Cases of male breast cancer were reported in eight families.

As a control, 155 unselected Italian healthy female blood donors were also analyzed. Median age was 43years in cases (range: 24–77) and 39years in controls (range: 20–69) ( P = 0.11). No allelic variant was identified for hsa-mir-17 and hsa-mir-369 , while a G to C allelic variation in mir-146a , corresponding to rs2910164 single-nucleotide polymorphism (SNP), was detected at comparable allele and genotype frequency in cases and controls ( Table I ). These frequencies, which conformed to the Hardy–Weinberg equilibrium ( P = 0.89 and P = 0.33 for cases and controls, respectively), were comparable also with those previously reported for normal European and Italian population ( 11 ; HapMap, http://hapmap.ncbi.nlm.nih.gov/ ; dbSNP, http://www.ncbi.nlm.nih.gov/projects/SNP ).

Table I.

mir-146a rs2910164 genotype and allele frequencies of BRCA1/BRCA2 -negative breast and ovarian cancer patients

Genotype	Cases (%)	Controls (%)	Allele	Cases (%)	Controls (%)
GG	60 (59)	90 (58)
GC	36 (36)	59 (38)	G	156 (77)	239 (77)
CC	5 (5)	6 (4)	C	46 (23)	71 (23)
Total	101	155	Total	202	310
P	0.87		P	1.00

Overall, these results rule out a role for genetic variations in these three miRNAs as major determinants in cancer predisposition of BRCA1/BRCA2 -negative patients. Instead, our study supports the hypothesis that mir-146a rs2910164 SNP may impact on the age of cancer onset, as suggested by Shen et al. ( 10 ).

In fact, when the non-parametric Wilcoxon test was applied to compare the age at diagnosis between genotype groups (the age at the first diagnosis was considered in the case of tumor multiplicity), it emerged that the patients with at least one mir-146a C allele (CC or GC) developed cancer at younger age compared with those of the GG group ( Table II ). The association between the genotype (GC + CC versus GG) and the median age at the diagnosis was confirmed also when the analysis was restricted to female cancers only ( Table II ).

Table II.

mir-146a genotypes and age of cancer onset a

Genotype	Whole series (95 females + 6 males)			Female breast + ovarian cancers			Female breast cancers only
	n	Median age (range)	P value	n	Median age (range)	P value	n	Median age (range)	P value
GG	60	46 (27–77)	Ref.	57	45 (27–77)	Ref.	53	45 (27–77)	Ref.
GC	36	41 (24–67)	0.023	33	40 (24–67)	0.037	30	39 (24–67)	0.033
CC	5	43 (26–49)	0.256	5	43 (26–49)	0.312	5	43 (26–49)	0.375
GC + CC	41	42 (24–67)	0.017	38	41 (24–67)	0.029	35	40 (24–67)	0.028

a

In the case of multiple tumors, age at the first diagnosis was considered.

To better visualize and appreciate the dynamics of cancer onset by genotype, we applied the Kaplan–Meier method. As shown in Figure 1 , the probability of developing a breast cancer at young age was higher in patients expressing at least one C allele (GC + CC versus GG, P = 0.024; GC versus GG, P = 0.026).

Fig. 1 .

Kaplan–Meier estimates by mir-146a genotypes in familial breast cancer women negative for BRCA1/BRCA2 mutations. The expression of at least one C allele (CC or CG) was associated to a higher probability of developing breast cancer at young age ( P = 0.024).

Although the G to C variation in the mir-146a precursor causes mispairing in the stem region, it has been shown that the miR-146a C allele displays a higher stability and higher binding affinity for the 3′-untranslated regions of BRCA1 compared with the G one ( 10 ). This results in higher efficiency in targeting BRCA1 messenger RNA for degradation and therefore may explain the ability of the C variant to promote early cancer onset even at a single dose.

Thus, in contrast to a recent report ( 11 ), our data support the hypothesis by Shen et al. of an association between the C allele of hsa-mir-146a and early cancer onset and prompt further investigations on the relevance of this polymorphism in early familial breast/ovarian tumor development.

Funding

Italian Ministry of Health; Associazione Italiana per la Ricerca sul Cancro - AIRC.

Conflict of Interest Statement: None declared.

Abbreviations

Abbreviations
 
• miRNA

  microRNA

 
• PCR

  polymerase chain reaction