Article Text

Download PDFPDF

Original article
Lynch syndrome caused by MLH1 mutations is associated with an increased risk of breast cancer: a cohort study
  1. Elaine F Harkness1,
  2. Emma Barrow2,
  3. Katy Newton2,
  4. Kate Green3,
  5. Tara Clancy3,
  6. Fiona Lalloo3,
  7. James Hill2,
  8. D Gareth Evans3
  1. 1Centre for Imaging Sciences, University of Manchester, Manchester, UK
  2. 2Department of General Surgery, University of Manchester and Central Manchester and Manchester Children's University Hospitals NHS Trust, Manchester, UK
  3. 3Manchester Centre for Genomic Medicine, St. Mary's Hospital, Manchester, UK
  1. Correspondence to Dr Elaine Harkness, Centre for Imaging Sciences, University of Manchester, Stopford Building, Oxford Road, Manchester, M13 9PT, UK; Elaine.F.Harkness{at}


Introduction Lynch syndrome is known to cause an increased risk of malignancies, including bowel and endometrial cancers. However, the risk of breast cancer associated with mutations in the mismatch repair (MMR) genes that cause Lynch syndrome is still unclear.

Materials and methods This study assesses the cumulative risk of breast cancer in 106 MLH1 and 118 MSH2 families. Families were referred on the basis of clinical criteria. Pedigree information was obtained, and tumour immunohistochemistry and microsatellite testing performed. Appropriate patients underwent sequencing and multiple ligation dependent probe amplification of all relevant exons of the MMR genes. Kaplan–Meier analysis of cumulative lifetime risk of breast cancer was made combining proven mutation carriers and their first-degree female relatives.

Results After allocation of mutation status, the cumulative risk of breast cancer to 70 years in MLH1 carriers was 18.6% (95% CI 11.3 to 25.9)). This is significantly higher than the cumulative risk for MSH2 which was 11.2% (95% CI 1.4 to 21.0) to age 70 years (p=0.014). The UK population risk is 7.5%–8% at the age of 70 years. Prospective analysis identified six breast cancers in 1120 years of follow-up with an OR of 3.41 (95% CI 1.53 to 7.59).

Discussions Female MLH1 carriers would appear to be at moderate risk of breast cancer and should be considered for breast screening at ages earlier than national screening programmes.

  • Hereditary Nonpolyposis Colorectal Cancer
  • DNA mismatch repair
  • Cancer: breast
  • Kaplan-Meier analysis

Statistics from


Lynch syndrome (previously hereditary non-polyposis colorectal cancer) is an inherited cancer predisposition syndrome caused by inactivation mutations of the DNA mismatch repair (MMR) genes: MLH1, MSH2, MSH6 and PMS2, and mutations in EPCAM (leading to impaired DNA repair through inactivation of MSH2). It is associated with an increased lifetime risk of colorectal cancers and a spectrum of extracolonic tumours, including endometrial, ovarian, upper urothelial, biliary, brain and small bowel malignancy. The inclusion of breast cancer within this syndrome is controversial.

Early studies on the risk of breast cancer in Lynch syndrome were based on individuals ascertained by differing clinical criteria1–3 (including, in some cases on the basis of prior breast cancer).4 Subsequent to the mapping of the MMR loci in 1993, several studies have examined the relative risk in MLH1 and MSH2 carriers. Scott et al5 found that breast cancer incidence was over-represented in MLH1 carriers, but not in MSH2 carriers (17 MLH1 families and 15 MSH2 families). This was contradicted by Vasen et al6 who did not find an increased risk for 187 MLH1 mutation carriers and 141 MSH2 mutation carriers, although the mean age at breast cancer diagnosis in carriers was young at 46 years. More recently, a prospective study by Win et al7 found an increased risk of breast cancer in 446 unaffected mutation carriers (MLH1 and MSH2 combined) over a median follow-up of 5 years. Win et al8 ,9 also demonstrated that women carrying MMR gene mutations were at increased risk of breast cancer following endometrial or colorectal cancer. Women with MSH2 mutations were more likely to develop breast cancer following colorectal cancer,8 but there was no significant difference between the genes following endometrial cancer, although the number of cancers was small.9

Engel et al10 also reported an increased risk of breast cancer in 2118 mutation carriers (806 MLH1, 1004 MSH2 and 308 MSH6) but there was no significant difference in risk between genes.

Additional evidence for the role of MMR mutation in the pathogenesis of breast cancers is provided by the presence of microsatellite instability (MSI) in breast tumour tissue from subjects with both MLH1 and MSH2 mutations.11–13 MSI is present in only a fraction of the sporadic breast cancers studied,14 and the occurrence of instability in breast tumours of MMR mutation carriers implies a distinct molecular genetic basis for their origin.

The Regional Genetics Service at St Mary's Hospital covers the population of the North West England. Two hundred and twenty-four Lynch syndrome families with proven pathological mutations in the MMR genes are known to this service. We report the cumulative lifetime risk of breast cancer for mutation carriers within this cohort.


Families are referred for genetic assessment based on clustering of cancers known to be associated with Lynch syndrome. Colorectal cancer is the most common of these. As these patients are not referred on the basis of breast cancers, the ascertainment bias should be limited. However, analysis of only the positive and obligate cases within the database will bias towards highly penetrant alleles, leading to an overestimation of cancer risk.

Women with a pathogenic mutation in MMR genes were identified using the departmental database for Lynch syndrome families. From the position in the pedigree, obligate carriers were also identified.

Kaplan–Meier curves for cumulative incidence of breast cancer were then generated for women for each gene in the following categories: (1) proven mutation and obligate carriers and (2) first-degree relatives (FDRs) of proven mutation carriers with unknown mutation status. The cumulative risk of breast cancer at 50, 60 and 70 years was assessed with 95% CIs. Kaplan–Meier curves were compared using the log rank test. Statistical analysis was performed in SPSS V.22.0.

To take into account any ascertainment bias and to improve the estimation of breast cancer risk within these families, a second analysis was carried out. In this analysis, carrier status was allocated to a proportion of the untested FDRs based on the following formula that was developed from a previous analysis of BRCA1/2 families:15

  • FDRs of a proven carrier with a Lynch syndrome-related cancer were considered carriers (we found only one instance (from 24 gastric cancers) of affected relatives with these cancers testing negative). The cancers included within the Lynch spectrum were colorectal, endometrial, ovarian, gastric, brain, biliary, small bowel and sebaceous adenocarcinoma.

  • A proportion of untested, unaffected FDRs with no cancers were assigned carrier status, based on the proportion of relatives who tested positive out of the total number of individuals actually tested for each age group (table 1), as previously described for BRCA1/2.15

  • Breast cancer is not currently considered part of the Lynch syndrome spectrum; therefore, the FDRs untested with breast cancer were not all assigned carrier status. This was allocated on the following premise:

    • Breast cancer has a population cumulative lifetime risk of 10%–12% and a 7.5%–8% risk to 70 years.16

    • The ratio of breast cancer risk to age 70 years generated from the initial analysis of proven carriers of each gene and their FDRs compared with population risks was used to allocate carrier status to FDRs of unknown status.

    • A person-years at risk analysis was carried out from date of family ascertainment with censoring at date of death or last known follow-up. Breast cancer incidence rates were compared with North West Cancer Intelligence Service age-specific incidence rates for the general female population in order to obtain expected numbers. The ORs were calculated based on the observed to expected ratios and are presented with 95% CIs.

Table 1

Proportion of unaffected female first-degree relatives testing positive and negative for family MLH1/MSH2 mutation at different ages


From the database of 2545 index cases and relatives, 356 women with a pathogenic mutation in MLH1, MSH2 or MSH6 were identified. In addition, a further 73 obligate carriers (due to their position in the pedigree in relation to relatives testing positive for a mutation) were identified. Therefore, in total, there were 157 MLH1, 219 MSH2 and 53 MSH6 mutation carriers and positive obligates. Mutation status was unknown for 206 MLH1, 262 MSH2 and 31 MSH6 female FDRs.

Initial analysis

The cumulative incidence of breast cancer in mutation carriers and positive obligates was calculated, censoring at age 70 years. The Kaplan–Meier graph is shown in figure 1A. Of the MLH1 mutation carriers, 15/157 (two bilateral) developed breast cancer, equating to a cumulative risk to 70 years of 17.4%. Only 5/219 MSH2 carriers developed breast cancer giving a lower cumulative lifetime risk of 4.9%. The difference was significant (log rank p=0.010). In the 53 MSH6 carriers, three developed breast cancer; due to the small numbers, no further detailed analysis was performed for MSH6. Cumulative incidence in FDRs of unknown mutation status was 15.8% to age 70 years for MLH1 (14/206 women (two bilateral)) and 11.8% for MSH2 (12/262). Combining both mutation carriers and FDRs of unknown mutation status gave cumulative risks to 70 years of 16.5% and 8.1% for MSH1 and MSH2 carriers, respectively (log rank for difference p=0.035). No breast cancer case tested negative for an MLH1 family mutation (three isolated cases tested positive) but the only woman with an isolated breast cancer (no other Lynch syndrome cancer) tested negative for their MSH2 family mutation.

Figure 1

Cumulative risk of breast cancer for MLH1 and MSH2 for (A) proven mutation carriers and (B) proven mutation carriers and first-degree relatives with assumed mutation status.

Second analysis (to establish penetrance)

  1. The cumulative risk for MSH2 carriers was similar to population risk; therefore, untested FDRs with breast cancer were allocated carrier status on a 1:1 basis sequentially by age.

  2. The cumulative risk for MLH1 carriers was twice the population risk. Mutation carriers had a risk of 2.22-fold the current upper estimate for population risk. For FDRs, 50% will not carry the mutation and will be at population risk. In order to generate a risk of 15.8% to 70 years in the mutation carriers a higher risk would be required. If we took a sample of 2000 women, 1000 would not carry the mutation and generate 75–80 cases of breast cancer by 70 years. In order to arrive at a 15.8% cumulative risk for both carriers and non-carriers combined (316 in 2000) a further 236–241 cases would have to occur in the mutation carriers, equivalent to 23.6%–24.1% or a relative risk of 2.95–3.21. Therefore untested FDRs with breast cancer were conservatively allocated carrier status on a 2:1 basis sequentially by age.

Following allocation of carrier status to the FDR untested group, an additional 230 (104 MLH1 and 126 MSH2) patients were entered into the analysis, giving, in total, 261 MLH1 carriers and 345 MSH2 carriers. Overall, the cumulative risk of breast cancer at age 70 years for mutation carriers was 13.2%. The revised Kaplan–Meier curve is shown in figure 1B, comparing the cumulative risks of MLH1 and MSH2. The risk to age 70 years for MLH1 was 18.6%, while that for MSH2 was 11.2%. The difference between MLH1 and MSH2 carriers was statistically significant (p=0.014). There were very few MSH6 carriers for any statistical inference to be made in this group. The life tables and risk per decade with 95% CIs are shown in table 2. The index cases were included in the analysis. Five of 15 MLH1 untested FDRs with breast cancer had other Lynch syndrome cancers compared with only 1/16 for MSH2.

Table 2

Cumulative risk for positive and negative carriers based on assumed status

Prospective analysis

Finally we carried out an assessment of breast cancer incidence in female MLH1 carriers from date of family ascertainment. In 1437 woman-years of follow-up (mean=7.6, median=6.8), six invasive breast cancers (five in MLH1 proven carriers) occurred in 112 mutation positive MLH1 carriers and 77 FDRs when 2.14 were expected (OR 2.81 95% CI 1.26 to 6.24). Confining the analysis to the 112 mutation carriers, there were five breast cancers in 861 woman-years of follow-up with only 1.55 expected (OR 3.23 95% CI 1.34 to 7.75). Finally, as most of the FDRs with unknown status would not carry the family mutation, we carried out a further analysis as previously described to allocate FDRs to mutation positive status. This reduced the total number of assumed carriers to 145 and the years of follow-up to 1120 with 1.76 cancers expected giving an OR of 3.41 (95% CI 1.53 to 7.59).


This study examines the risk of breast cancer in individuals with proven MMR mutations and their FDRs. The study demonstrates that even after correction for potential ascertainment bias, MLH1 mutation carriers have a significantly higher cumulative lifetime risk of breast cancer than MSH2 carriers (log rank p=0.014). Within the assumed status MLH1 mutation carriers, there were four cases of bilateral breast cancer compared with no cases within the MSH2 carriers—further evidence of an increased risk. The risk of breast cancer for MSH2 carriers was similar to population risk. In a further prospective element of the present study, we show a significant threefold relative risk of breast cancer in MLH1 carriers. Prospective analysis provides the most compelling evidence for any increased risk.7

Combining mutations in all three genes, the CIs for the overall cumulative risk to age 70 years includes population risk. This may explain the previous contradictory data from studies before stratification by gene, but does not explain the similar risks for MSH2 and MLH1 mutation carriers in the Dutch study.6 However, the more recent study by Engel et al,10 using Dutch data and data from Germany, found a significantly increased risk of breast cancer for all mutation carriers compared with the general population, but no difference in risk between MSH1 and MSH2 carriers. We also confirmed an increased risk in untested FDRs. Women with a similar genetic background who develop colorectal cancer may be protected against breast cancer by having less female hormone (hormone replacement therapy is known to protect against colon cancer but increase breast cancer risk), although this is not supported by the Win et al8 data. Confining risk to proven carriers many of whom will have been ascertained through development of colorectal cancer may conceal an overall increase in risk. Indeed, Win et al8 found that following colorectal cancer carriers of the MMR gene mutations had a 1.76 increased risk of developing breast cancer. The reverse of this situation has been suggested for colon cancer in BRCA1 carriers.17

These findings have significant implications for the counselling and screening of women with Lynch syndrome secondary to MLH1 mutation. The risk of breast cancer before 50 years of age at 6%–7% is considerably higher than the 2% population risk.16 The risk to 70 years of 15.8%–18.6% is about double the population risk of 8%.16 The 95% lower CIs for the assumed status allocation do not encompass the current population risks of 3% at 50 years and 11.3% at 70 years despite the fact that the time period of observation includes the previous 40 years when breast cancer risks were lower. In the UK, under National Institute for Health and Care Excellence clinical guideline 41/164,16 ,18 these patients would fall into the ‘moderate risk’ category. As such, they could be offered annual mammographic surveillance from the age of 40 years. This is 7–10 years earlier than the entry age for the UK National Health Service breast screening programme for those at population risk.

In North America, annual mammographic surveillance begins at the age of 40 years for those with population risk of breast cancer, although recent guidance suggested changing this to 50 years. For those with increased risk, it is suggested that annual mammograms begin at the age of 30 years, that shorter surveillance intervals be used (eg, 6-monthly) and that alternative screening modalities such as ultrasound and MRI be considered.19 This may be necessary in view of the possible risks of extra radiation involved in annual mammographic screening at a younger age in women with mutations in MMR genes. Further studies on the safety of radiation exposure in MLH1 carriers may be needed before widespread implementation of extra mammography.

This study, in combination with the study by Scott et al,5 provides further evidence that MLH1 mutations are associated with an increased risk of breast cancer. It is important that these women are counselled appropriately and that breast screening is considered at an earlier age in line with local guidance.


The authors thank Leah Robinson and Waleed Alduaij for their previous work on the dataset.



  • Contributors EFH was involved in the data analysis and interpretation, and writing of the manuscript. DGE designed the study and drafted the manuscript. KG was responsible for the study database. EB, KN, TC, JH and FL were involved in data collection. All authors commented on previous versions of the manuscript and approved the final version.

  • Competing interests None declared.

  • Ethics approval Central Manchester Research Ethics Committee.

  • Provenance and peer review Not commissioned; externally peer reviewed.

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.