Introduction

Lynch syndrome, an autosomal dominant hereditary cancer syndrome, most often results from a germline mutation in one of the four DNA mismatch repair (MMR) genes: MLH1, PMS2, MSH2, or MSH6. In rare instances, it may also be caused by a germline mutation in the EPCAM gene, which leads to inactivation of MSH2 protein [1, 2]. Inheriting any of these mutations is associated with an increased risk of developing certain types of cancer, and usually with an earlier age of onset compared with sporadic tumors [1, 3,4,5,6]. The two most common tumor types are colon (53–82% lifetime risk) and endometrial cancer (25–60% lifetime risk), with endometrial cancer presenting as the first malignancy in ~50% of women with Lynch syndrome [1, 3]. Identification of patients with Lynch syndrome is critical as these individuals and their family members may benefit from genetic counseling and appropriate surveillance for cancer prevention or early detection [1, 3,4,5,6].

Currently, universal screening for Lynch syndrome has been adopted at many institutions for all patients with endometrial cancer. Testing algorithms may include MMR immunohistochemistry for MLH1, PMS2, MSH2, and MSH6 expression, and/or PCR testing for microsatellite instability (MSI) in tumoral tissue, followed by genetic counseling and germline genetic testing of selected patients. However, establishing the diagnosis of Lynch syndrome can be still challenging, as up to half of endometrial or colon cancer cases with MMR protein loss and absence of MLH1 promoter hypermethylation have no detectable pathogenic germline mutation in the MMR genes or EPCAM. This group of patients has been termed having “Lynch-like syndrome” and their appropriate clinical management remains problematic [4]. Sequencing of tumor DNA has been recently shown to resolve this uncertainty in up to 70% of such cases by identifying somatic mutations in MMR genes—two pathogenic mutations, or one pathogenic mutation with loss of heterozygosity [7, 8].

Recent studies identified loss of MMR protein expression specific to the known germline mutation in normal nonneoplastic colonic crypts of Lynch syndrome patients [9, 10]. In addition, the same finding was observed in one patient with “Lynch-like syndrome,” suggesting that presence of MMR-deficient crypt foci may be a novel indicator of Lynch syndrome [10]. However, this phenomenon has not yet been systematically evaluated in endometrial tissues. We present, to our knowledge, the first comprehensive analysis of MMR protein expression pattern in correlation with the germline MMR gene mutations in nonneoplastic endometrium of Lynch syndrome patients.

Methods

Patients with known Lynch syndrome and available normal endometrial tissues were identified retrospectively in our departmental archives. Clinical information and results of the germline genetic testing were collected from the patients’ medical records. Additional cases were identified to serve as controls in three groups: (1) normal endometrial tissue adjacent to sporadic MMR-intact endometrial carcinoma; (2) normal endometrial tissue adjacent to sporadic MMR-deficient endometrial carcinoma with MLH1 promoter hypermethylation (confirmed by methylation specific multiplex PCR); and (3) normal endometrial tissue from patients undergoing hysterectomy for benign indications without clinical suspicion of Lynch syndrome. Hematoxylin–eosin stained slides of the endometrial specimens were retrieved from the archives and were reviewed to select a block containing the most amount of benign endometrial glandular epithelium.

Four MMR immunostains were performed on every Lynch syndrome and control case using primary monoclonal antibodies against MLH1 (clone M1, Ventana, Tucson, AZ), PMS2 (clone EPR3947, Cell Marque, Rocklin, CA), MSH2 (clone G219-1129, Ventana), and MSH6 (clone 44, Ventana), according to the manufacturers’ instructions. The immunostained slides were reviewed by at least two of the authors and assessed for presence of background internal positive control and any loss of nuclear expression in the benign endometrial glands (in all Lynch syndrome and control cases) and MMR expression pattern within the endometrial carcinoma (in one Lynch syndrome case and in control groups 1 and 2). In Lynch syndrome cases with no loss of MMR staining on the initial block, two additional tissue blocks were selected and stained for the MMR protein corresponding to the patient’s known germline mutation.

Fisher’s exact test and Student’s t test were used for statistical analysis and p values of <0.05 were considered statistically significant.

Results

We identified 27 female patients with a known diagnosis of Lynch syndrome who had undergone hysterectomy (n = 19) or endometrial curettage/biopsy (n = 8). The diagnosis of Lynch syndrome was established by germline mutation testing: most patients harbored a pathogenic mutation in MSH2 (12 of 27, 44%), followed by MSH6 (7 of 27, 26%), PMS2 (6 of 27, 22%), and MLH1 (2 of 27, 7%). The patients’ age at the time of procedure ranged between 31 and 61 years (mean: 45.6 years) (Table 1). Five patients (19%) had a personal history of gastrointestinal cancer, four of them of the colorectum and one of the ampulla. In one patient the hysterectomy was performed for a known diagnosis of endometrioid endometrial adenocarcinoma, while the other 18 hysterectomies were prophylactic, two of which harbored incidental complex atypical endometrial hyperplasia. All endometrial curettings/biopsies were performed as part of a routine surveillance and showed benign endometrial tissue. The patient with known endometrial carcinoma and the two patients with incidental complex atypical hyperplasia all displayed the expected pattern of MMR protein loss in the lesional tissue corresponding to their germline mutations (PMS2 in the patient with endometrial carcinoma, and MSH2 in both patients with atypical hyperplasia).

Table 1 Mismatch repair protein expression in nonneoplastic endometrial glands by immunohistochemistry.
Full size table

Loss of MMR protein expression was seen in singles or clusters of morphologically normal, nonneoplastic endometrial glands on the initial tissue sections in 15 of the 27 Lynch syndrome cases (56%) (Table 2, cases #1–15). Of the remaining 12 cases, additional tissue blocks were available in 10 cases (all hysterectomy specimens), in which two additional tissue blocks were selected for each case and stained for only the MMR protein corresponding to the patient’s known germline mutation. Staining of additional tissue blocks increased the yield of detection of MMR-deficient benign endometrial glands to 19 of 27 (70%) Lynch syndrome cases (Tables 1 and 2). Loss of MMR protein expression was specific to the known germline mutation in each patient: two cases of MLH1 mutation (loss of MLH1 and PMS2 expression), four cases of PMS2 mutation (loss of PMS2 expression), ten cases of MSH2 mutation (loss of MSH2 and MSH6 expression), and three cases of MSH6 mutation (loss of MSH6 expression) (Figs. 1–4). In cases with mutations in MLH1 or MSH2, loss of staining within the same glands were also seen in their paired heterodimers, PMS2 or MSH6, respectively (Figs. 1 and 3). In one case (case #11, Table 2), MMR IHC was performed on both the endometrial curettage and subsequent hysterectomy specimen, and identical loss of MSH2 and MSH6 staining was seen in clusters of benign endometrial glands in both specimens. The two Lynch syndrome cases with atypical hyperplasia (cases #8 and #9 in Table 2) both displayed loss of MMR immunostaining in the area of hyperplasia and in clusters of benign endometrial glands immediately adjacent to and also several low magnification fields away from the hyperplastic glands. The only one Lynch syndrome case with endometrial adenocarcinoma (case #17 in Table 2) did not show loss of MMR expression in nonneoplastic glands on the initial sections. However, one of the two additional nonneoplastic sections—from the lower uterine segment, away from the endometrial carcinoma—showed loss of PMS2 immunostaining. The loss of MMR expression involved single glands in six cases and clusters of glands in 13 cases. No statistically significant correlation was observed between the two groups of Lynch syndrome patients (MMR intact versus MMR loss) with regards to the affected germline MMR genes or the patients’ age (p > 0.05) (Table 3).

Table 2 Lynch syndrome patients with loss of mismatch repair protein expression in benign endometrial glands by immunohistochemistry.
Full size table
Fig. 1: Lynch syndrome case #2 (Table 2), MLH1 germline mutation.
figure 1

Prophylactic hysterectomy with morphologically unremarkable secretory endometrium (a, b). A single gland (marked with * on all panels) shows loss of both MLH1 (c) and PMS2 (d) expression. (a, c, d: original magnification ×100, b: original magnification ×200).

Full size image
Fig. 2: Lynch syndrome case #4 (Table 2), PMS2 germline mutation.
figure 2

Routine surveillance endometrial biopsy shows early secretory phase endometrium (a, b). A cluster of glands is identified with retained MLH1 (c) and loss of PMS2 (d) expression. (a, c, d: original magnification ×100, b: original magnification ×200).

Full size image
Fig. 3: Lynch syndrome case #11 (Table 2), MSH2 germline mutation.
figure 3

Routine surveillance endometrial curettage shows secretory phase endometrium (a). PMS2 expression is retained (b), while paired loss of MSH2 (c) and MSH6 expression (d) is seen in the same cluster of glands. (All images at original magnification ×100).

Full size image
Fig. 4: Lynch syndrome case #15 (Table 2), MSH6 germline mutation.
figure 4

Routine surveillance endometrial biopsy shows proliferative phase endometrium (a, b). MSH2 expression is retained (c) while loss of MSH6 expression (d) is noted in the same cluster of benign glands. (a, c, d: original magnification ×100, b: original magnification ×200).

Full size image
Table 3 Mismatch repair protein expression pattern in nonneoplastic endometrium in Lynch syndrome cases.
Full size table

A total of 56 cases comprised the control group: 27 sporadic MMR-intact endometrial carcinomas (18 FIGO grade 1 or 2 endometrioid carcinomas, 3 mixed endometrioid and serous carcinomas, 4 serous carcinomas, 1 clear cell carcinoma, and 1 carcinosarcoma), 9 sporadic MMR-deficient endometrial carcinomas with loss of MLH1 and PMS2 expression and MLH1 promoter hypermethylation (8 FIGO grade 1 or 2 endometrioid carcinoma, 1 dedifferentiated carcinoma), and 20 hysterectomy specimens for benign indications (e.g., leiomyomas or adenomyosis) (Table 1). All cases in the three control groups showed retained (intact) expression of MMR markers in benign endometrial glands.

In addition, we identified two patients harboring germline variants of unknown significance (VUS) in MMR genes MSH2 c.2293G > A (p.A7657) and PMS2 c.2149G > A (p.V717M). Neither one of the patients had a personal history of colon cancer. However, both patients had family history of ovarian and breast cancer, and underwent prophylactic hysterectomy due to the perceived uncertainty regarding the risk of endometrial cancer and appropriate clinical management. No histologic evidence of atypical hyperplasia or carcinoma was identified and MMR immunohistochemistry showed no loss of expression in benign endometrial glandular epithelium in either cases.

Discussion

Our study demonstrated loss of germline mutation-specific MMR protein expression by immunohistochemistry in nonneoplastic endometrial glands in 70% of patients with Lynch syndrome. In contrast, all normal endometria in the three control groups—adjacent to MMR-intact, or MMR-deficient MLH1 hypermethylated sporadic endometrial carcinoma, and in benign hysterectomy specimens—showed retained, intact staining patterns with all four MMR markers. These findings may provide important insights into the pathogenesis of Lynch syndrome-associated endometrial cancer, and also raise the possibility that MMR protein expression analysis of preneoplastic or nonneoplastic tissues may be useful in separating true Lynch syndrome from sporadic “Lynch-like” cases.

In the gastrointestinal tract, loss of MMR protein expression has been recently reported in 25–35% of nonneoplastic colonic and small bowel crypts of Lynch syndrome patients, a subset of which also demonstrated MSI by PCR [9,10,11]. In addition, one of the patients with “Lynch-like syndrome” also showed loss of MSH2 staining in nonneoplastic colonic crypts [10]. The sensitivity of detection was increased by deeper level sectioning of tissue blocks to evaluate additional mucosal surface area [9,10,11]. MMR deficiency—by either IHC, PCR, or both—has also been identified by other studies in over 70% of colonic adenomas from Lynch syndrome patients [12,13,14].

The literature is more limited on the prevalence and potential significance of MMR deficiency in precancerous lesions of the endometrium. Identical loss of MMR protein expression by IHC and/or MSI by PCR have been described in areas of endometrial hyperplasia (both with and without atypia) adjacent to endometrial carcinoma in a small number of Lynch syndrome patients [15,16,17]. Of note, one of these studies found a higher frequency of MSI and an earlier average age of onset of carcinoma in MSH2 mutation carriers compared with other MMR gene mutation carriers, raising the possibility that MSH2 mutation may indicate a more rapid rate of tumor progression [15]. In unselected patient populations in two tissue microarray-based studies MMR protein loss was identified in 4.5% [18] and in 20% [19] of atypical endometrial hyperplasia cases, although the case number was much smaller in the latter study. Nieminen et al. analyzed 110 endometrial samples collected over several years of routine cancer surveillance from 54 women with Lynch syndrome, all of whom subsequently developed endometrial hyperplasia or endometrial carcinoma [20]. MMR IHC was performed only on a subset of their Lynch syndrome endometrial samples (n = 49) and showed decreased MMR protein expression corresponding to the patients’ known germline mutation in 100% of complex hyperplasia without atypia, in 92% of complex hyperplasia with atypia, in 40% of simple hyperplasia, and in 7% (2 of 29) of normal endometrial samples [20]. The germline mutation affected MLH1 in both patients in the latter group, although specific details of the decreased MMR staining patterns were not presented. Most recently Lucas et al. studied MMR protein expression in atypical hyperplasia/endometrioid intraepithelial neoplasia from 63 patients with MMR-deficient endometrial carcinoma, including 14 patients with genetically confirmed Lynch syndrome [21]. Background atypical hyperplasia was present in 8 of 14 Lynch syndrome patients, all of which demonstrated the same loss of MMR protein expression as the corresponding carcinoma. Interestingly, normal-appearing background benign endometrium was also present in seven of their cases, but showed no loss of MMR staining [21].

The details of molecular pathogenesis of endometrial cancer in Lynch syndrome are not yet fully understood. In the gastrointestinal tract, Lynch syndrome is not associated with an increased rate of adenoma formation, and Lynch syndrome-associated colorectal cancer is traditionally thought to arise through accelerated progression of preformed (MMR-proficient) adenomas [22]. An alternative pathogenetic pathway has also been recently proposed, suggesting that morphologically normal MMR-deficient colonic crypts could give rise to carcinoma, either directly or through an MMR-deficient adenoma phase [12, 23]. Similarly, it is conceivable that MMR-deficient nonneoplastic endometrial glands may represent the initial step in endometrial carcinogenesis in Lynch syndrome patients either leading to the development of atypical hyperplasia or directly progressing to carcinoma. Prior studies demonstrated that Lynch syndrome-associated endometrial carcinoma coexists with complex atypical hyperplasia in ~40% of cases, similar to the rate observed in sporadic endometrial cancer, supporting that there is a continuum of disease progression through complex atypical hyperplasia in Lynch syndrome patients [24, 25]. Interestingly, however, mutations in PIK3CA, KRAS, and CTNNB1 were found to be less frequent in Lynch syndrome-associated atypical hyperplasia and endometrial carcinoma compared with sporadic cases, while PTEN mutations and loss of PTEN expression were seen at a similar frequency between the two groups, suggesting that in the presence of MMR deficiency loss of PTEN function may be sufficient for carcinogenesis [24]. PTEN inactivation has also been observed by immunohistochemistry in up to 43% of histologically normal-appearing proliferative endometria from an unselected patient population, and these PTEN-null glands have been hypothesized to represent the initial phase of a multi-step carcinogenesis [26, 27]. Progression would require accumulation of additional genetic abnormalities and the risk is modulated by both hormonal and nonhormonal mechanisms [26, 28]. The significance of the hormonal milieu, i.e., unopposed estrogen effect, in endometrial carcinogenesis is well known and prior studies have shown significant regression of PTEN-null glands following progestin therapy [26, 29, 30]. The possibility of hormonal inactivation of MMR genes have not been extensively studied, but given the specificity of our results—loss of MMR protein expression matched the patients’ known germline mutations, and none of the patients in the control groups displayed MMR loss—it is unlikely to be a significant contributing factor. In addition, loss of MMR protein expression was seen in Lynch syndrome patients with a wide age range (31–61 years) and various menstrual cycle phase (inactive, proliferative, interval, and secretory phase; see Table 2).

Other potential mechanisms of MMR deficiency have also been reported in colorectal cancer and in adjacent precursor lesions. For example, CpG island methylation can result in silencing of tumor suppressor genes and hMLH1 MMR gene and thereby promote tumorigenesis [31,32,33,34]. Promoter methylation was also found to be frequent in sporadic endometrial endometrioid carcinomas and in adjacent histologically normal endometria, although hMLH1 was not included in these studies [35, 36]. We did not observe loss of MLH1 or other MMR protein expression in our control cohort of normal endometrial tissues (including controls adjacent to sporadic MMR-deficient MLH1 hypermethylated tumors), and the only two Lynch syndrome patients with loss of MLH1 expression in benign glands had known germline MLH1 mutations. In addition, cell differentiation and cell cycle phase are known to affect MMR gene expression, and MMR protein immunohistochemical expression is generally stronger in actively proliferating tissues compared with inactive ones (e.g., proliferative phase versus atrophic endometrium) [37, 38]. Similarly, we observed that the intensity of MMR immunohistochemical staining in inactive/atrophic endometrial tissues was generally weaker (both in the Lynch syndrome and in the control groups) compared with proliferative or secretory phase endometria. However, no complete loss of staining was identified in any of the control cases and the loss of staining among Lynch syndrome cases was specific to the patient’s known germline mutation in each case. Several studies have also described aberrant patterns or loss of MMR expression in colorectal cancer and in other tumor types as a result of prior chemo- and/or radiation therapy [39, 40]. The possibility of treatment-related MMR-alteration can be excluded in our Lynch syndrome cohort, as only one of the patients was diagnosed with endometrial carcinoma and she did not receive chemotherapy or radiation prior to the hysterectomy.

Our findings may also have potential implications for Lynch syndrome screening. Many institutions (including ours) recently adopted universal screening of all newly diagnosed endometrial cancer cases regardless of the patient’s age. Most screening algorithms recommend MMR protein immunohistochemistry (for all four—MLH1, PMS2, MSH2, and MSH6—or for at least two—PMS2 and MSH6—markers) as the initial step, followed by MLH1 promoter methylation analysis in cases with combined loss of MLH1/PMS2 [41, 42]. Tumors with MLH1 promoter hypermethylation are likely sporadic and do not require additional genetic workup unless there is a strong clinical suspicion for Lynch syndrome. Patients with loss of both MSH2 and MSH6, or isolated loss of PMS2 or MSH6 expression in their tumors should be referred to genetic counseling and targeted germline testing. MSI testing by PCR has been reported to have a 90% concordance rate with MMR IHC and has been integrated into the testing algorithm either as an initial step, co-testing with IHC to maximize the detection rate, or as a secondary assay for cases with indeterminate/equivocal MMR staining patterns or intact MMR staining and strong clinical suspicion for Lynch syndrome [43, 44]. Universal screening maximizes the detection of Lynch syndrome in endometrial and colorectal cancer patients but it also presents a significant clinical challenge as over 50% of patients with MMR protein loss and absence of MLH1 promoter hypermethylation lack detectable germline MMR gene or EPCAM alterations [4, 41, 45]. A proportion of these cases may harbor somatic MMR gene mutations within the tumor, while the remaining “Lynch-like syndrome” patients may have true Lynch syndrome harboring germline MMR gene alterations that are not yet known or are undetectable by currently available methods [46]. In addition, germline testing may identify MMR VUS. Patients harboring germline MMR gene VUS and those with “Lynch-like syndrome” are faced with uncertainty regarding the need or frequency of lifelong cancer surveillance (both colorectal and endometrial) and screening of their asymptomatic family members for Lynch syndrome.

In this study, we report presence of MMR-deficient, morphologically normal, nonneoplastic endometrial glands in 70% of Lynch syndrome cases. While the sensitivity of this phenomenon for Lynch syndrome is only 70% (30% of cases would be false negative if used for screening purposes), the specificity was 100% as it was not observed in any of the control cases. Our detection rate increased significantly (from 56 to 70%) by staining additional tissue blocks, and it is conceivable that further increase in the number of stained blocks and deeper level sections would result in even higher sensitivity. In contrast to the previous observations in the gastrointestinal tract, which only included resection specimens for known colorectal or small bowel carcinoma [9,10,11], we were able to demonstrate MMR-deficient benign endometrial glands in both hysterectomy and endometrial curettage/biopsy specimens and all but one of our Lynch syndrome cases were prophylactic surgeries or routine surveillance biopsies with no known endometrial cancer. MMR-deficient benign endometrial glands were present in patients as young as 31 years of age, suggesting that they may precede development of atypical hyperplasia or carcinoma by several months or years. The loss of MMR protein expression was specific to the patients’ known germline mutations in all cases, and in patients with MLH1 or MSH2 germline mutations identical loss of protein expression was also seen in the paired heterodimers, PMS2 and MSH6, respectively. Although we did not observe a statistically significant correlation between the affected germline MMR gene and presence of MMR-deficient benign endometrial glands, it was most frequently identified in MSH2 (10 of 12, 83%) and MLH1 (2 of 2, 100%) mutation carriers.

Interpretation of MMR staining results in the Lynch syndrome group was generally straightforward, as the endometrial glands were in proliferative or secretory phase in most cases and both the glandular epithelium and adjacent stroma showed nuclear staining serving as internal controls. Control cases with adjacent carcinoma were typically from older patients showing inactive/atrophic background endometrium, which often showed only weak nuclear MMR expression. However, staining intensity in glandular epithelium was always compared with internal control (surrounding normal stromal tissue and inflammatory cells), and MMR staining was interpreted only in areas with satisfactory internal control staining.

The distribution of germline MMR gene mutations in our study population is different from the known frequency of those previously described among Lynch syndrome patients. Most prior studies reported MLH1 and MSH2 mutations to be the most common, accounting for ~60–80% of all Lynch syndrome cases, while the minority of patients (5–10%) have mutations in PMS2 or MSH6 [47,48,49]. In contrast, the most common germline mutation among our Lynch syndrome cases was in MSH2 (44%), followed by MSH6 (26%) and PMS2 (22%), while only 7% of patients harbored germline MLH1 mutations. A potential explanation may include a selection bias, as most previously reported mutation frequencies were calculated based on Lynch syndrome patient groups with a known diagnosis of colorectal cancer, while only 19% of our patients had a prior diagnosis of colorectal carcinoma. In addition, our relatively small sample size may have also contributed to these differences. Our study also included two cases with germline VUS, involving MMR genes MSH2 and PMS2, both of which showed intact MMR protein expression in benign endometrial glandular epithelium by immunohistochemistry. We did not identify any “Lynch syndrome-like” patients in our study cohorts; however, future studies would be necessary to determine the significance of our findings in those cases.

In summary, we present a novel finding of frequent loss of germline mutation-specific MMR protein expression in benign, morphologically normal endometrial glands in Lynch syndrome patients. We hypothesize, that presence of MMR-deficient nonneoplastic endometrial glands may represent an early detectable marker of Lynch syndrome, and may be further explored as a potentially useful screening tool in endometrial curettings/biopsies from patients with suspected Lynch syndrome. Furthermore, this observation may also have significant pathogenetic implications, raising the possibility—at least in a subset of Lynch syndrome patients—of endometrial carcinogenesis directly from morphologically normal glands without a stepwise progression through a preneoplastic (atypical hyperplasia) phase.