Introduction
Germline mutation in one of four genes with DNA mismatch repair (MMR) function,
MLH1, MSH2, MSH6, and
PMS2, causes susceptibility to cancers of multiple organs known as Lynch syndrome (LS) [
1]. Among all cancers, those of the colon and rectum, endometrium, small bowel, ureter and renal pelvis have the highest relative risk compared to the general population, which is why they are considered to be part of the LS tumor spectrum according to the Amsterdam criteria II [
2]. Among extracolonic tumors accepted as LS spectrum tumors when the Amsterdam II criteria were formulated, cancer of the renal pelvis had the lowest relative risk (14) [
3].
Breast cancer is the most common cancer among women worldwide [
4]. It is presently not included in the LS tumor spectrum because its incidence has not been found to be elevated in LS patients, whether carriers of
MLH1 or
MSH2 mutation [
5‐
8] or
MSH6 mutation [
9]. A few deviating reports exist. A study on Australian LS patients revealed a significantly increased standardized incidence ratio (14.77) for breast cancer in
MLH1, but not
MSH2 mutation carriers [
10]. The standardized incidence ratio for breast cancer was also elevated (3.95) in a prospective investigation of a cohort of 446 unaffected carriers of
MLH1, MSH2, MSH6, and
PMS2 mutation over a median follow-up of five years [
11]. Furthermore, among Brazilian families meeting the most stringent clinical (Amsterdam I) criteria for LS [
12] but with no mutation data available, breast cancer was the most frequent extracolonic cancer in women, even exceeding the frequency of endometrial cancer [
13].
As an additional approach to decide whether breast cancer is a LS spectrum tumor or not, breast cancers arising in LS patients have been studied for microsatellite instability (MSI) or MMR protein expression since deficient MMR is a hallmark of LS. As observed originally, the absence of MMR defects appeared to exclude breast cancer as an integral tumor of LS [
14] whereas subsequent studies report the occurrence of MMR defects in breast carcinomas from MMR gene mutation carriers with frequencies ranging between 44% and 75% based on MSI and/or aberrant MMR protein expression [
8,
15‐
19]. Frequencies of MMR defects reported in the latter studies exceed those independently observed for sporadic breast carcinomas (0 to 20%) [
20,
21] suggesting that MMR deficiency is an important factor in breast cancer development in LS.
Given the conflicting results from previous studies, we collected all available LS-associated breast carcinomas through the nation-wide Hereditary Colorectal Cancer Registry of Finland. Molecular profiles in breast carcinomas from mutation carriers were compared to those from proven non-carriers (phenocopies) from mutation-positive families and to sporadic breast carcinomas from the same population; additionally, a large collection of other LS-associated tumors from mutation carriers was included in the analysis. Molecularly, breast carcinomas from LS mutation carriers were divided into two subgroups paralleled by some clinical differences (notably, different ages at onset) as will be described below.
Discussion
Genetic and clinical heterogeneity characterize LS and manifest themselves in the wide spectrum of cancers that can be found in LS families [
1,
34]. We set out to address whether breast carcinoma is part of the LS tumor spectrum based on molecular characteristics of the individuals and tumors. To our knowledge, our study is the first to compare breast carcinomas from mutation carriers to those from proven non-carriers (phenocopies) and to sporadic breast carcinomas from the same population. This setting provides a reliable reference for the background rates of the alterations examined. For example, many cancers that occur in LS are also common in the general population, reflecting the combined effect of low-penetrance susceptibility alleles and environmental factors [
35]. Moreover, the pattern of organ involvement has changed over time including, for example, a relative decline in stomach cancer and an increase in colorectal cancer incidence in more recent generations [
36] and may vary according to the ethnic or geographic origin [
13,
37], suggesting that the LS tumor spectrum is sensitive to environmental influences. The present utilization of representative sets of other tumors from carriers of the same mutations additionally makes it possible to recognize the important role that the tissue of origin may play as a mediator of the effects of MMR deficiency [
38] or epigenetic dysregulation [
39]. Finally, previous molecular studies on LS-associated breast carcinoma have mainly focused on the MMR status; we additionally evaluated the epigenetic profiles of the tumors.
Our previous epidemiological investigation of MMR gene mutation carriers from Finnish LS families [
5] arrived at a standardized incidence ratio of 1.4 for breast cancer compared to the general population, suggesting that breast cancer incidence is not elevated in Finnish LS families. This finding is in agreement with most reports from other populations (see Introduction). In the present study, breast carcinoma was diagnosed at 56 years of age in carriers and 54 years in non-carriers, which is comparable to sporadic breast carcinomas from our population (59 years) [
33]. In the available literature, the average age of breast carcinoma diagnosis in MMR gene mutation carriers varies between 46 years and 66 years, including both early-onset series [
6,
18,
40] and series with the age at onset similar to the general population [
8,
17]. Interestingly, the age at diagnosis was significantly lower for MMR-deficient- (53 years) versus MMR-proficient breast carcinomas (61 years) among mutation carriers from our study. Combined with the normal life-time risk of breast cancer in our LS families, the finding is compatible with the idea that MMR defects might preferentially be involved in breast cancer progression rather than initiation [
6]. The breast carcinomas we studied from MMR gene mutation carriers were predominantly ductal, ER- and PR-positive and HER2-negative, resembling breast carcinomas from non-carriers (Table
1) and those from the unselected Finnish population [
33,
41]. Similar to our findings, ductal histology predominates in published reports on LS-associated breast carcinomas [
17‐
19]. While the hormone receptor or HER2 status has seldom been specified in previous papers, a single study [
19] found predominant hormone receptor negativity among MMR-deficient breast cancers from MMR gene mutation carriers; furthermore, there was no difference in the age at onset between MMR-deficient and MMR-proficient breast cancers. Apart from possible population-specific differences, reasons for the conflicting results relative to ours are unknown.
Apparent selectivity in tumor spectrum despite ubiquitous expression is a puzzle shared by a majority of high-penetrance susceptibility genes, including the MMR genes in LS. In theory, the LS tumor spectrum could be influenced by the predisposing gene and mutation (for example, carriers of
MSH2 mutation may have a higher risk of various extracolonic tumors compared to
MLH1 mutation carriers [
6]). MMR gene dosage, which affects the extent and degree of the MMR defect, is also important, both constitutionally (for example, biallelic germline mutations are associated with a distinct tumor spectrum characterized by hematological and central nervous system malignancies, [
42]) and somatically (for example, DNA damage signaling requires a higher dosage of MMR protein than DNA mismatch repair [
43] and inactivation of both alleles of a MMR gene may, therefore, not always be necessary in tumor cells). Different target genes are rate-limiting in different cancers and their susceptibility to mutations and selective advantage conferred by the mutations may vary (for example,
TGFBRII and
PTEN both contain coding repeats which are structurally prone to frameshift mutations in the context of deficient MMR, but
TGFBRII is mainly involved in gastrointestinal cancers and
PTEN in endometrial cancer [
44]. Finally, different organs may be differently exposed to exogenous (for example, dietary carcinogens) or endogenous agents (for example, hormone-induced oxidative stress) and removal of such damage may depend on functional MMR [
45,
46].
In our study, 65% (13/20) of breast carcinomas from MMR gene mutation carriers showed loss of the MMR protein corresponding to the germline mutation and 35% (8/23) had high-degree MSI. In comparison with other tumors from MMR gene mutation carriers, the percentage of breast carcinomas with MMR protein inactivation was consistently lower and the proportion with MSI-H in the lower range (Figure
3A). The fact that
MSH6 mutations were overrepresented among patients with breast carcinomas (7/23, 30%) versus other tumors (4/105, 4%) might have some influence on the results since the percentage of MMR-deficient breast carcinomas was lower among
MSH6 than
MSH2 or
MLH1 mutation carriers (Table
2). Importantly, we observed that a given MMR gene mutation could be associated with MMR proficient breast carcinomas in some carriers and MMR-deficient breast carcinomas in other carriers of the same mutation (see Table
2 and Figure
1 for examples), ruling out the possibility that normal protein expression or microsatellite-stability was a general characteristic of these mutations. We conclude that breast carcinomas from MMR gene mutation carriers conformed to other LS tumors in that IHC analysis of MMR protein expression was more sensitive than MSI analysis to detect a MMR defect but deviated from the remaining tumors in that even IHC failed to detect a MMR defect in a considerable proportion (35%) arguing against biallelic inactivation in those cases. Even so, the proportion of MMR-deficient breast carcinomas was significantly higher compared to breast carcinomas from non-carriers (Table
1) suggesting that MMR defects do play a role in breast tumorigenesis in MMR gene mutation carriers. Furthermore, as discussed above, the possible role of MMR gene malfunction in processes other than MMR cannot be completely excluded in the case of MMR-proficient tumors.
The role of growth-regulatory target genes was addressed through studies of promoter methylation, which is known to be an important mechanism of inactivation of the TSGs examined [
25,
47]. The average number of methylated TSGs out of 24 was in the upper range in breast carcinomas among all tumors from MMR gene mutation carriers (Figure
3B). The five most frequently methylated TSGs (
RASSF1, APC, CDH13, GSTP1, and
CDKN2B) were the same in all groups of breast cancer (Figure
2) and, with the exception of
CDKN2B, were involved in other tumors from LS patients, too, although with variable frequencies (Figure
3B).
CDKN2B promoter methylation may be a particular characteristic of breast tumorigenesis as also supported by observations from other groups [
48].
Competing interests
The authors declare that they have no competing interests.
Authors' contributions
JEL performed all methylation studies for breast tumors. AG conducted IHC analyses for breast tumors. AG, TTN, and JEL were responsible for MSI analyses. TK, WA-R, RP and MF contributed to histological analyses of tissue specimens and evaluated IHC results. J-PM, HJ, MA and KA diagnosed and recruited the Lynch syndrome cases and TK provided specimens from the sporadic cases. PP, J-PM, TK, and JEL conceived the study and drafted the manuscript. All authors read and approved the final manuscript.