Introduction

Lynch syndrome (LS), also known as hereditary non-polyposis colon cancer syndrome (HNPCC), is one of the most common forms of inherited predisposition to colorectal cancer (CRC), representing 1–7 % of all CRC [1–3]. More recently the prevalence of LS has been determined to be one out of every 35 CRC patients (2.8; 95 % CI, 2.1–3.8 %) [4]. LS is characterized by an earlier age at diagnosis of CRC (median age of 44) [5, 6] than that in sporadic CRC (70 years) (http://seer.cancer.gov/csr/1975_2008/results_single/sect_01_table.11_2pgs.pdf). Individuals with LS have 80 % probability of developing CRC by the age of 65 and are at elevated risk of developing a second primary CRC [7]. In addition, they have increased risk of developing various extra-colonic malignancies. For example, affected women have a 50–60 % risk of developing endometrial cancer. Cancer at other primary sites including stomach, ovaries, small intestine, urological tract, pancreas, brain and skin [1, 8, 9] has also been documented in families with LS. Among those extra-colonic malignancies, endometrial, renal pelvis/ureter, and small intestinal cancers have the highest relative risk and are considered as part of LS spectrum [10]. Also part of the LS spectrum are sebaceous gland neoplasms and keratoacanthomas along with low grade visceral malignancies that are classified as a subtype of LS called Muir-Torre syndrome [11, 12].

Germline mutations in DNA mismatch repair (MMR) genes are responsible for LS. The main genes involved are MLH1, MSH2, MSH6, PMS1 and PMS2 [10, 13, 14]. Studies revealed some evidence for a relationship between the involved MMR gene and the cancer spectrum. For example, MSH2 mutation carriers appear to have a higher risk of developing extra-colonic cancer than individuals with MLH1 mutation [7, 15, 16]. In addition, the frequency and the types of mutations are reported to differ in different geographical areas, therefore, the frequency of extra-colonic cancer associated with LS may be different in different populations [17–19].

To better characterize LS and further study the cancer spectrum related to LS, in this study we have evaluated the frequency of extra-colonic cancer in 176 unrelated LS families enrolled in a hospital based LS registry.

Methods

In this descriptive analysis, we studied LS patients and family members with confirmed pathogenic DNA mismatch repair (MMR) mutations. These were all participants who had consented to participate in an ongoing IRB approved study “Molecular Genetics of HNPCC” and were recruited between September 1994 and June 2011 at MD Anderson Cancer Center. Probands were largely recruited from the gastroenterology and gynecologic oncology clinics at MD Anderson Cancer Center and also through the genetic counselors at the Clinical Cancer Genetics Clinic. Family members were consented and enrolled through the probands. The subjects included both cancer affected and cancer free MMR gene mutation carriers. Demographic characteristics and cancer history were self reported through a self administered structured questionnaire. Whenever possible, cancer diagnoses were further confirmed from medical records and pathology reports.

Statistical analysis

Demographic characteristics and MMR gene mutated were compared between the cancer affected and cancer free mutation carriers using simple χ2 statistics. The distribution of colonic versus extracolonic cancers was compared by sex, race, age and MMR mutation. Furthermore, the distribution of cancers and the median age at diagnosis for each class of cancers in the study sample were explored. Additionally, cancer specific standardized incidence ratios (SIRs) and 95 % confidence intervals were computed to determine the excess cancers in MMR mutation carriers compared to noncarriers, and in the subset of mutation positive relatives. We excluded the probands in the subset analysis for cancer specific SIRs to overcome the significant bias associated with probands, since probands are ascertained largely because of being cancer affected, resulting in a high cancer rate in probands. SIRs were computed as the ratio of the observed number of cancer cases to the expected number using the Cohort Analysis for Genetic Epidemiology (CAGE) software program [20]. The expected number of cancers was determined using sex, age and birth cohort specific rates from the Surveillance Epidemiology and End Results (SEER) data. The CAGE program uses first primary cancers from SEER to compute the expected number of cancers so analyses were limited to calculating SIRs for the first primary cancer of each site other than non-melanoma skin cancer, which is excluded in SEER.

Results

The study included 368 MMR mutation carriers from 176 families, of which 59 % were women. The subjects were predominantly non-Hispanic white (83.4 %). The ages ranged from 18 to 93 years (median 49 years). Roughly 30 % of the mutation carriers were cancer free. There were no significant differences between the cancer affected and cancer free mutation carriers by sex, race, age (at index cancer diagnosis for CRC cases and at last contact for cancer free subjects), MMR gene mutated or the mutation type (Table 1). There were a total of 165 probands (for whom we had data), of which 142 (86 %) had cancer, and 203 mutation carrying family members, of which 118 (58 %) were cancer cases. The number of family members per family ranged from 1 to 15. The number of cancers per person ranged between 1 and 7, with roughly half (51.6 %) diagnosed with only 1 cancer. The cancers included in the analysis were diagnosed between 1948 and 2010 (median 1997). A total of 504 cancers were recorded in the 368 MMR gene mutation carriers. There were 236 (46.8 %) colorectal and 268 (53.2 %) extracolonic cancers overall. However, CRC was the most common index cancer, n = 151 (58.1 %) as compared to 109 (41.9 %) index extracolonic cancers. Table 2 outlines the comparison between the colorectal and extracolonic cancers by sex, race, MMR gene mutated, type of MMR gene mutation and age at diagnosis of any cancer. These results are also presented as Fig.1a–e. Men had fewer extracolonic cancers than colorectal (44.6 vs. 55.4 %), whereas women had more extracolonic cancers than colorectal (58.9 vs. 41.1 %), (P = 0.002). MLH1 mutation carriers had more colorectal cancers but there were more extracolonic than colorectal cancers in MSH2 and MSH6 and PMS2 mutation carriers (difference not statistically significant). The mean age at diagnosis for extracolonic cancers was older than for colorectal, 49.1 versus 44.8 years (P ≤ 0.001). Distribution of median age at diagnosis and the minimum and maximum age at diagnosis for each of the cancers is shown in Fig. 1f. and distribution and types of all cancers seen in this cohort is described in Figs. 2a and b.

Table 1 Demographic characteristics of MMR mutation carriers
Full size table
Table 2 Distribution of colorectal and extracolonic cancers
Full size table
Fig. 1
figure 1

Distribution of index cancers by, a all index cancers, b sex, c MMR gene, d age, e race, f median age at diagnosis (lines represent minimum and maximum ages at diagnosis)

Full size image
Fig. 2
figure 2

Distribution of all cancers by, a all cancers, b sex

Full size image

Many rare cancers presented as index cancers as outlined in Table 3. The index tumor was not a LS related tumor in 26 (10 %) cancer patients.

Table 3 Distribution of index cancers
Full size table

SIR values were significantly higher for cancers of the colon, rectum, endometrium, ovaries, kidneys, bladder, stomach and small intestine. However, for the non-Lynch syndrome related cancers like breast, prostate, melanoma, cervix, lung, sinus cavity and thyroid, the SIR values were not significant (Table 4).

Table 4 Cancer specific standardized incidence ratios for all MMR mutation carriers and for MMR mutation positive relatives (excluding probands)
Full size table

Discussion

In this report of cancers affecting 368 MMR gene mutation carriers in a hospital based Lynch syndrome registry, we analyzed the tumor spectrum to determine the frequency and distribution of index and multiple cancers. We compared the distribution of colonic and extracolonic cancers by sex, race, age and MMR mutation, and our study confirmed previously reported findings that extracolonic cancers are more common in women perhaps because of their predisposition to endometrial and ovarian cancers [1] and that extracolonic cancers are more likely to be associated with MSH2 and MSH6 mutation carriers [21].

Surveillance of these high risk patients entails not just screening for colorectal and endometrial cancer but also cancers at other sites. The Lynch syndrome spectrum of cancers besides colorectal and endometrial includes stomach, ovaries, small intestine, urological tract, pancreas, brain and skin. Overall in our sample, ~10 % of the index tumors were located at sites that are not considered to be part of the Lynch syndrome spectrum. For example, there were 5 patients that presented with breast cancer as the primary cancer and 3 patients that presented with prostate cancer. Whether breast cancer is part of Lynch syndrome tumors is controversial with evidence both in support of [22, 23] and against it [24, 25, 26]. Similarly there have been reports suggesting that prostate cancer may be a part of the Lynch syndrome tumor spectrum [8, 27, 28]. Although there were a significant number of breast and prostate cancer cases in our study, the SIR for these cancers were not significant as these are relatively common cancers in the general population. However, seeing that these tumors were detected in MMR mutation carriers, and were seen in much younger women (median age of onset for breast cancer: 36 years) and men (median age of onset for prostate cancer: 65 years) than expected, may suggest a high index of suspicion for considering Lynch syndrome in early onset patients presenting with these cancers.

Several rare non-Lynch syndrome index cancers were recorded in this population. These included thyroid (n = 3, 1.15 %), cervix/vagina (n = 3, 1.15 %), melanoma (n = 3, 1.15 %) and 1 case each of thymoma, carcinoma of the sinus cavity, and adenocarcinoma of the lung. In addition, there was 1 index brain tumor and 4 patients presented with metastatic disease (lung, liver, unknown primary). With MD Anderson being a tertiary care hospital, it is possible that these cases may not represent the population spectrum of LS cancers.

The strengths of our study include a large sample size of MMR gene mutation carriers from 176 unrelated families and detailed recording of cancer history.

In conclusion, we found that the distribution of cancers in patients in our hospital based Lynch syndrome registry was similar to what has been previously reported. However, the findings regarding index cancers seen in our study suggest that a significant number of Lynch syndrome patients can present with primary tumors that are not part of the Lynch syndrome spectrum.