Lynch Syndrome–Associated Breast Cancers: Clinicopathologic Characteristics of a Case Series from the Colon Cancer Family Registry

Defects in autosomal dominant genes are suspected to be responsible for up to 10% of all breast cancers (1, 2). The most commonly affected known genes are BRCA1 and BRCA2, whereas other genes such as p53, PTEN, STK11/LKB1, and CDH1 are implicated in fewer cases (3, 4). There are, in addition, a significant number of cases of breast cancer that present at an earlier than usual age of onset and with a conspicuous family history for which the causative gene(s) has not been identified (5).

Lynch syndrome, or hereditary nonpolyposis colorectal cancer (6), was first identified nearly a century ago as familial clustering of cancers, particularly of the colon, small intestine, stomach, endometrium, upper urinary tract, and sebaceous tumors of the skin. Approximately 80% of Lynch syndrome–associated cancers are attributable to defects in the DNA mismatch repair (MMR) genes MLH1 and MSH2, with the majority of the remaining cases occurring in carriers of mutations in MSH6 and, to a lesser extent, PMS2 (7). The consideration of breast cancer as a spectrum tumor in Lynch syndrome has been controversial, with evidence for and against constituting a rigorous debate over time. An extensive study published in 2002 excluded breast cancer as part of the Lynch syndrome spectrum of tumors (8). Another study, while demonstrating no increased risk for developing mammary cancers, found that tumors in known carriers presented at an earlier age (9). This led de Leeuw et al. to postulate that, whereas MMR deficiency does not in itself initiate breast tumors, the increased rate of mutation accelerates their progress leading to an earlier presentation (10). In contrast to these observations, Scott et al. reported a significant 15-fold increased risk of breast cancer in MLH1 mutation carriers (but not MSH2 carriers; ref. 11), and a study of Brazilian Lynch syndrome families showed an increased incidence of breast cancer equal in incidence to endometrial cancer cases (12), although this study used clinical criteria to define Lynch syndrome. In more recent times, both case reports (13, 14) and statistical studies (15, 16) have shown that MMR-deficient breast cancers can and do arise in mutation carriers.

Immunohistochemical screening of tumors for deficiency in MMR proteins is currently the most efficacious method for the recognition of Lynch syndrome (17). Immunohistochemistry allows not only the identification of potential Lynch syndrome patients, but also by its pattern of staining, the most likely causative gene. This is important because tracking down a mutation in MMR genes is not a trivial exercise. Immunohistochemistry is generally applied to recognized Lynch syndrome spectrum tumors such as those from the colorectum, endometrium, ovary, stomach, and urothelial tissues. This has also become increasingly important as family size decreases and screening becomes more widespread, lessening the power of clinical criteria to detect Lynch syndrome, and highlighting the need to increase the potential sources of tissue that may be used for diagnosis. Breast cancers with both microsatellite instability (MSI) and/or immunohistochemical absence of MMR proteins have been reported by many authors (13, 14, 18–23), although many of these reports contain small numbers of cases. In this study, we sought to establish the frequency with which breast cancers occurring in mutation carriers for Lynch syndrome display MSI as a consequence of loss of DNA MMR proteins, and to examine the clinicopathologic features of such tumors. This study represents the largest series of breast cancers with MMR deficiency reported to date, and attests to the utility of breast tissue as a diagnostic sample in suspected Lynch syndrome when colon or endometrial tissue is unavailable (13).

One hundred and seven cases of breast cancer (arising in 102 females and 2 males) from 90 colorectal cancer families were identified from the Colorectal Cancer Family Registry (Colon CFR), a National Cancer Institute–supported consortium established in 1997 to create a comprehensive collaborative infrastructure for interdisciplinary studies of the genetic and molecular epidemiology of colorectal cancer (see detailed information about the registry at the CFR web site;²²

Families were selected for study in which (a) both breast and colon cancer co-occurred, with at least one breast cancer regardless of age at diagnosis; (b) families met either modified Amsterdam criteria (ACII), or had at least one early-onset (<50 y) colorectal cancer; and (c) breast tissue was available within the biospecimen repository for MMR testing, thereby limiting the number of families which could be analyzed. Comprehensive cancer histories and tissue were available for 90 families recruited through the Australasian Colorectal Family Registry and the Mayo Clinic Cooperative Family Registry for Colon Cancer Studies. The majority of families (n = 86) were enrolled in the Colon CFR on the basis of a strong family history of colorectal cancers compatible with Lynch syndrome, and the remaining four families were identified following enrollment of participants with early-onset (<50 y) colorectal cancer.

Fifty-four of 90 (60%) families met modified ACII, and the remainder had multiple cancers including at least one early-onset (<50 y) colorectal cancer per family. In 13 cases (13%), the breast cancer patient was also affected with early-onset colorectal cancer, 58 individuals (56%) were first-degree relatives of an individual with early-onset colorectal cancer, 25 individuals (24%) were second-degree relatives, and the remaining 8 cases (8%) were more distantly related. All reported tumors (breast and other sites) were verified by either examination of the original histopathology material or histopathology reports for all affected kindred members where possible. In three cases, bilateral metachronous tumors were available for testing. In no instance was there evidence of familial adenomatous polyposis. No deleterious mutations in the breast cancer susceptibility genes BRCA1 and BRCA2 were detected in the small minority of families (n = 7) tested. As the Colon CFR recruits families on the basis of colorectal cancer, it is unlikely that families enrolled in this registry would be of the type of configuration that would trigger BRCA1 testing.

One consultant histopathologist (M.C. Cummings) reviewed material from 104 of the 107 cases of breast cancer to confirm diagnosis and score histopathologic features. Clinicopathologic data for the remaining three cases was abstracted from histopathology reports. Noting the following features: tumor location, size, primary histologic type, tumor grade (using the Nottingham modification of the Bloom-Richardson system; ref. 25), tumor margin, confluent necrosis, calcification, presence of tumor-infiltrating and peritumoral lymphocytes, presence of in situ carcinoma and atypical ductal hyperplasia and axillary lymph node status. Although note was also made of the steroid hormone receptor status as originally reported, these markers were re assessed by immunohistochemistry in our laboratory owing to the large number of incomplete reports. HER2/neu status was extracted, where available, from original laboratory reports.

Immunohistochemistry for DNA MMR proteins was done as previously described (26). A subset of tumors was stained for estrogen receptor (ER), progesterone receptor (PR), and p53. Paraffin sections (4 μm) were subjected to heat-induced epitope retrieval in High pH Target Retrieval Solution (Dako Corporation) for ER and PR, and Reveal Decloaker Solution (BioCare Medical) for p53. Sections were stained with rabbit monoclonal anti-human ER (clone SP1) or rabbit monoclonal anti-human PR (clone SP2) at 1:2,500 or mouse monoclonal anti-p53 antibody at 1:100 followed by the EnVision Plus Mouse HRP detection system (Dako) for p53, or MACH3 Rabbit HRP polymer kit (BioCare Medical) for ER and PR. The proportion of positive cancer cell staining was graded as follows: 0 (negative), <10% (1+), 11% to 25% (2+), 26% to 50% (3+), 51% to 75% (4+), and >75% (5+). Tumors were scored as positive when there was strong expression in >10% tumor cells. Histologically normal breast epithelium present within tumor blocks served as the positive control for ER and PR, and sections from known p53-overexpressing colorectal cancers were used as positive controls when staining for p53 in breast cancers.

Tumors were assessed for MSI using a panel of 10 microsatellite markers and classified as MSI-high if ≥30% of the markers showed instability, MSI-low if one or more markers but <30% of all markers showed instability, and microsatellite stable if no marker exhibited instability, as has been previously reported (17). Only cases with five or more evaluable markers were considered. Methylation of the MLH1 promoter was detected using the MethyLight Assay as has been recently described (27).

The somatic T>A mutation at nucleotide 1799 causing the V600E mutation in the BRAF gene was determined using a fluorescent allele-specific PCR assay. Briefly, 20 to 50 ng of DNA, extracted from formalin-fixed, paraffin-embedded tumor tissue, was amplified in a 25 μL reaction containing 100 nmol/L each of allele-specific primers tagged with differing fluorophores [mutant primer (F1): 6-Fam, 5′-CAGTGATTTTGGTCTAGCTTCAGA-3′; wild-type primer (F2): NED, 5′-TGATTTTGGTCTAGCTACAGT-3′; and a common reverse primer (REV), 5′-CTCAATTCTTACCATCCACAAAATG-3′], together with 2.5 units of Taq polymerase (Eppendorf), 1× buffer, and 200 μmol/L of deoxynucleotide triphosphates. The cycling conditions consisted of an initial denaturation at 95°C for 2 min followed by 35 cycles at 94°C for 30 s, 59°C for 30 s, and 65°C for 30 s then a final extension at 65°C for 10 min. After amplification, 1 μL of the PCR product was added to an 8.7-μL mix of HiDi formamide and ROX Genescan 500 size marker (Applied Biosystems). The mutant allele (A1799) primer generated a PCR product of 97 bp, 3 bp larger than the wild-type PCR product after separation on an ABI 3100 genetic analyzer. GeneMarker (SoftGenetics) software was used to identify the different size and fluorescent allele PCR products. Positive and negative controls were run in each experiment and 10% of samples were replicated with 100% concordance.

DNA (10 ng) was amplified in 25 μL reactions using HotMaster Taq and buffer (Eppendorf) with 20 pmol of each primer. Cycling protocols were applied according to previously established conditions for each primer set and amplicon, verified to selectively amplify the target amplicon only. PCR products were cleaned using Millipore Montage PCR96 Cleanup Plates (Millipore). Cleaned PCR product (1 μL) was used in a 12 μL sequencing reaction utilizing the BigDye Terminator v3.1 reagents and protocol (Applied Biosystems) and 2 pmol of primer. Sequencing product was cleaned with the DyeEx 96 Kit (Qiagen) using the recommended protocol. The product was dried, resuspended in HiDi Formamide (Applied Biosystems), and analyzed on an ABI PRISM 3100 (Applied Biosystems). Bidirectional sequencing was done throughout. Results were compared with reference sequence NC_000003.10 (genomic) and NM_000249.2 (cDNA) for MLH1, and NC_000002.10 (genomic) and NM_000251.1 (cDNA) for MSH2, NC_000002.11 (genomic) and NM_000179.2 (cDNA) for MSH6, and NC_000007.13 (genomic) and NM_000535.3 (cDNA) for PMS2. Multiplex ligation-dependent probe amplification (MLPA) was used to detect large exonic deletions and duplications in the four MMR genes, using the Salsa MLPA P003 and P248 kits for MLH1 and MSH2, and the P008 kit for MSH6 and PMS2 (MRC-Holland) according to the protocols of the manufacturer.

Statistical analysis was carried out using Statistical Package for Social Sciences (version 17.0). Contingency tables were assessed using χ² or Fisher's exact test as appropriate. Differences between means were assessed using a t test to ensure equality of the variance in groups using probability plots and an F test. P < 0.05 was considered significant. Sensitivity and specificity calculations were done using VassarStats.²³

Of 90 families with breast and colorectal cancer, 53 families (59%) were classified as Lynch syndrome on the basis of a germ line mutation (n = 52) or multiple MSH2/MSH6-deficient tumors within the family (n = 1). Table 1 shows the distribution of families among the four causative MMR genes (MLH1, MSH2, MSH6, and PMS2). In the 37 remaining families, Lynch syndrome could be excluded as no evidence of MMR deficiency from immunohistochemistry or MSI testing was found, nor were any MMR mutations identified where tested. In 30 families, Lynch syndrome was excluded on the basis of immunohistochemistry and MSI testing of several family members with no evidence of MMR deficiency. The remaining seven families showed multiple MLH1-deficient colorectal and/or endometrial cancers associated with MLH1 methylation and/or somatic BRAF mutation (colorectal cancer only) with no evidence of germ line mutations in MLH1 either by direct sequencing or MLPA. Only one of these seven families met modified ACII. None of the 61 breast cancers in this study, which could be analyzed, showed the V600E activating mutation in BRAF.

Abnormal immunostaining for MMR protein was observed for 18 of 107 breast cancers (17%; Fig. 1; Supplementary Figs. S1 and S2). In 12 cases, tumors showed loss of MSH2 and MSH6 proteins, in five cases, MLH1 and PMS2 were absent, and in one case, only MSH6 was absent. MSI testing was done for 89 breast cancers and was concordant with immunohistochemistry results in 85 cases (96%). All but one of the MMR-deficient breast cancers arose in families meeting ACII. Of the 18 participants whose breast cancers showed loss of one or more MMR proteins on immunohistochemistry, 16 tested positive for germ line mutations in the DNA MMR genes MLH1, MSH2, or MSH6 consistent with both their tumor immunodeficiency as well as their respective family mutation, while another individual who was deceased was found to be an obligate carrier of her family mutation (Table 2). The remaining case demonstrating immunohistochemical loss of MSH2 and MSH6 arose in a family in which no MSH2 mutation has been identified to date but which has three affected kindred members wherein their tumors show commensurate loss of MMR proteins.

Overall, 18 of 35 known MMR mutation carriers with breast cancer (51%) produced a breast cancer that was MMR deficient. Of the 89 breast cancers with normal immunohistochemistry, 13 arose in individuals from families in which MLH1 mutations were identified, 20 in individuals with a family MSH2 mutation, 5 with MSH6 mutations, and 3 with PMS2. Of these 41, 17 individuals tested carried the family mutation, suggesting that only a proportion of breast cancers in mutation carriers are associated with MMR deficiency.

Ten of the 18 individuals showing loss of one or more MMR proteins in their breast cancers had other primary tumors as summarized in Table 3. In 3 of the 10 cases, breast cancer was the first diagnosed malignancy, preceding the second cancer by between 2 and 27 years. In the other patients, the breast cancers were diagnosed between 1 and 42 years after the first cancer. In all cases but one (a meningioma), the non–breast cancers tested showed the same pattern of MMR protein deficiency as the breast tumors, a finding which supports the premise of this report, i.e., that the breast cancers in mutation carriers that are MMR deficient are likely to have developed in association with the germ line mutation carried. Importantly, in eight cases of breast cancer in a proven mutation carrier, breast cancer was the only cancer documented.

There was no statistical difference in age of presentation between the MMR-deficient breast cancers (mean, 57.5 ± 8.1 years; range, 43.4-75.0 years) and MMR-intact breast cancers (mean, 55.8 ± 11.9 years; range, 36.1-86.7 years; P = 0.56), nor between the MMR-deficient breast cancer group and MMR-intact known mutation carriers (57.1 ± 12.0 years; range, 36.1-80.5 years; P = 0.90). Similarly, no difference in mean age of presentation was observed between the five breast cancer cases which were MMR-deficient in MLH1 germ line mutation carriers and the 12 cases occurring in MSH2 carriers (58.0 versus 57.5 years; P = 0.90). The average age of the 11 individuals with the primary or only cancer being a breast cancer with MMR deficiency was 53.7 ± 6.0 years.

In 104 breast cancers that underwent pathology review, only histologic differences for invasive breast cancers were compared, with 12 cases of ductal carcinoma in situ excluded from analysis. Specifically, MMR-deficient invasive breast cancers (n = 16) were more likely to be ER- and PR-negative (P = 0.031 and P = 0.022, respectively), have peritumoral lymphocytes (P = 0.015), have confluent necrosis (P = 0.002), have growth in solid sheets (P < 0.001), and to have a higher mitotic rate (P = 0.002) when compared with MMR-proficient breast cancers (n = 79). In addition, MMR-deficient breast cancers less frequently had contiguous in situ disease (P = 0.038). No statistically significant association was seen between MMR status and tumor type, size, lymphovascular invasion, node status, prominent eosinophilic nucleoli, or tumoral calcification (Table 4; Supplementary Table S1). No statistical differences were observed for clinicopathologic features between the MMR-proficient invasive breast tumors arising in known carriers of germ line MMR gene mutations and tumors from the non–Lynch syndrome group (data not shown), and therefore, all MMR-intact tumors were considered together as the reference group for comparison with MMR-deficient invasive cancers. There were, however, significant differences in growth in solid sheets (P = 0.002), and the presence of pushing margins (P = 0.042), and confluent necrosis (P = 0.017) and residual carcinoma in situ (P = 0.004) between MMR-deficient and -intact invasive cancers among proven carriers of MMR gene germ line mutations (Table 5).

Four tumors displayed typical BRCA1 histologic phenotype (characterized by high grade, high mitotic index, pushing margin, growth in solid sheets, and the presence of lymphocytic infiltrate and tumor necrosis; ref. 28), and two of these tumors (50%) showed loss of MSH2 and MSH6.

The two cases of ductal carcinoma in situ which exhibited loss of MMR expression were both of solid type, but there was no significant difference overall between DCIS type (cribriform, solid, papillary, or clinging) and MMR expression when including in situ disease accompanied by an invasive component (P = 0.45; data not shown). Lobular carcinoma in situ was present in eight cases accompanying invasive disease. There was no statistically significant difference between MMR-deficient and -proficient breast cancers and overexpression of p53 (P = 0.39).

There was a trend for individuals with a MMR-deficient breast cancer to have also developed an early-onset colorectal cancer, or was a first-degree relative of someone so affected (n = 15) when compared with individuals with a MMR-proficient breast cancer having more distantly related cases of early-onset colorectal cancer (n = 3; P = 0.11, 21% versus 9%, respectively). Of the 13 cases in which the same individual had both early-onset colorectal cancer and breast cancer, 5 (39%) showed MMR deficiency. There was no statistical difference between degree of kinship between the breast and early-onset colorectal cancer patients within individual families and whether the pedigree satisfied the modified ACII (P = 0.17).

In the present study, we have examined the incidence of MMR deficiency occurring in breast cancers from families with a history of early-onset (<50 years) colorectal cancer and breast cancer occurring at any age. Of the 104 individuals investigated, 35 were found to harbor deleterious mutations in one of the DNA MMR genes, and of these, 18 individuals (51%) were found to have loss of MMR expression consistent with their respective germ line mutations. This study follows previous reports in which several groups have examined breast cancers with varying panels of markers to assess the extent of instability in tumors from this site. Many, however, have yielded disappointing results in which little or no instability could be detected in the majority of tumors (22, 29–32), although these studies were not specifically designed to detect Lynch syndrome. In 1996, Risinger and colleagues described breast cancer occurring in Lynch syndrome kindreds that showed high-level MSI. On the basis of this, it was suggested that breast cancer might be included in the tumor spectrum of Lynch syndrome II (21). A subsequent study failed to show an increased risk for breast cancer in Lynch syndrome, and further suggested that sporadic breast cancer with MMR deficiency may be exceedingly rare (8). However, reports of breast cancer with MMR-deficient phenotypes continued to be presented (9, 10, 33), with more targeted studies subsequently demonstrating that MSI is indeed a common feature of breast cancers occurring in known mutation carriers, being found in up to 60% of cases (10, 34, 35). In our study, in which MSI and immunohistochemistry results were highly correlated, we returned a figure of 51% for the proportion of breast cancers (which included a male breast cancer) arising in MMR mutation carriers that showed MMR deficiency, commensurate with this previously reported figure.

A study by Vasen and coworkers of 200 putative Lynch syndrome families has shown that although breast cancer occurs at an early age in Lynch syndrome, there is no elevated risk per se (9). This suggested that MMR deficiency may accelerate tumorigenesis in breast cancers which occur in Lynch syndrome mutation carriers but is unlikely to be the initiating event (10). We found no significant difference between MMR-proficient and MMR-deficient breast cancers when age of onset was analyzed. The mean age at diagnosis was 57 years in our study. Vasen et al. reported a mean age of 46 years for seven cases of breast cancer arising in mutation carriers (9), or 50 years as reported by both Stupart et al., who examined the incidence of breast cancers in women carrying in common a single mutation in MLH1 (c. C1528T; ref. 36), and Jensen et al., for a series of breast cancers arising in 20 mutation carriers reported recently (35). The numbers of cases in all studies were, however, small, and recruitment biases in different studies might account for any differences.

Medullary carcinomas of the breast show morphologic similarities to the MSI-high colorectal tumors (poor differentiation and lymphocytic infiltrate) and a proportion of these are MSI-high (37). Many of the studies to date investigating the issue of breast carcinomas arising in the setting of Lynch syndrome have been anecdotal case reports of one or two patients who have shown loss of appropriate MMR proteins by immunohistochemistry and/or high levels of MSI in tumors arising in proven mutation carriers (19, 38, 39). With such small numbers in any given study, it has been difficult to determine whether there is an “MSI” phenotype associated with such cancers, although Yee et al. described higher levels of MSI in lobular carcinomas (39%) than in infiltrating ductal cancers (13.5%; ref. 40). The issue has been made all the more difficult in that, whereas some breast cancers in mutation carriers have appropriate MMR protein loss with resultant MSI, there are a commensurate number of reports of breast tumors arising in proven carriers which have competent MMR (8, 39, 41). In this report, we found that breast cancers with proven MMR deficiency are significantly more likely than those with proficient MMR to show hormone receptor negativity, poor differentiation, a solid growth pattern, lymphocytic infiltrate, high mitotic rate, confluent necrosis, and vesicular tumor nuclei. In common with the study by Jensen et al., we found that ductal carcinoma not otherwise specified (NOS) was the predominant histotype (35), with no evidence of overrepresentation of specific types such as medullary or invasive lobular carcinoma.

Several of these features, including poor differentiation and lymphocytic infiltration, are also reported features of Lynch syndrome colorectal cancers (42, 43), and a dense lymphocytic infiltrate has been previously shown in a breast cancer case arising in a Lynch syndrome mutation carrier (14), and more recently, in a larger series of MMR-deficient breast cancers arising in mutation carriers in which three of the six tumors were reported to have both tumor-infiltrating lymphocytes and peritumoral lymphocytes (35).

The presence of breast cancer in Lynch syndrome has been reported to be overrepresented in Lynch syndrome families with MLH1 mutations (11, 15). Rarer causes of Lynch syndrome such as germ line mutations in MSH6 have also been associated with synchronous breast and colon cancers (44). However, due to our limited study design, we are unable to offer any comments regarding whether or not the risk of breast cancer is increased in Lynch syndrome mutation carriers.

Familial aggregation of cancers from different anatomic sites has been previously documented (45–47). Such clustering may arise from shared environmental risk factors common to the cancers, inherited defects in cancer susceptibility genes or interaction between the two. A series of large studies using the Swedish Family-Cancer Database, concluded that in the absence of known, strong environmental risk factors, most of the familial aggregation that is observed is likely due to genetic factors that increase the risk of cancer at more than one site (48). In addition, it has become increasingly apparent that most inherited cancer susceptibilities confer a risk for cancer at a range of sites, suggesting that mechanisms of carcinogenesis are shared by different tissues. This is true of Lynch syndrome, in which a defect in DNA MMR clearly confers an increased risk for cancers of the colorectum, endometrium, ovary, stomach, bladder and renal pelvis. As described in multiple previous reports, we found gene-appropriate MMR deficiency to be also readily demonstrable in approximately half of the breast cancers arising in Lynch syndrome mutation carriers using MMR protein immunohistochemistry, thereby confirming breast cancer tissue as a valid screening option for Lynch syndrome diagnosis. A caveat to testing breast cancers for Lynch syndrome however is that, although in this study we showed that loss of MMR by immunohistochemistry is 100% specific for Lynch syndrome, the sensitivity for detecting a mutation carrier is only 51.4%. This compares with a much higher sensitivity for colorectal cancers in which MMR-proficient phenocopies are relatively rare in mutation carriers.

Furthermore, we found that MMR-deficient tumors show certain histologic features significantly more often than would be expected thus increasing confidence in the selection and use of particular breast cancers for this purpose. It is worth noting that, of four cancers that displayed the “BRCA1” histologic phenotype, two showed loss of MMR proteins, suggesting that screening of such cases for Lynch syndrome might be considered where BRCA1 mutation testing proved fruitless. The finding that breast carcinoma was the only malignancy reported for half of the women with MMR-deficient breast cancers is consistent with the findings of Jensen et al., who reported no other cancer types in three of seven such cases (35).

No potential conflicts of interest were disclosed.

The authors are grateful to the many pathology laboratories involved for the supply of extra archived breast tissue for analysis. Thanks are due to Erika Pavluk, Judi Maskiell, Leanne Prior, and Maggie Angelakos for data and pedigree retrieval, as well as to the members of the families who gave significant time and effort to the contribution of data. Joanne Young is a Cancer Council Queensland Senior Research Fellow.

Grant Support: National Cancer Institute, NIH under RFA no. CA-95-011, and through cooperative agreements with the Australasian Colorectal Cancer Family Registry (U01 CA097735) and the Mayo Clinic Cooperative Family Registry for Colon Cancer Studies (U01 CA074800).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement inaccordance with 18 U.S.C. Section 1734 solely to indicate this fact.