Identification of possible Lynch syndrome in endometrial carcinomas at a public hospital in South Africa

Abstract

Endometrial carcinomas are common malignancies worldwide. Microsatellite instability is often identified in endometrioid endometrial carcinomas (EECs) and in most Lynch syndrome (LS) associated tumours. The authors aimed to identify the number of EECs at a South African public hospital, using the four-mismatch repair immunohistochemical stains, which may suggest possible LS, for the period 2009–2015. Following ethical clearance, 145 cases of archived EEC underwent immunohistochemical testing for MLH1, MSH2, MSH6 and PMS2. Cases demonstrating loss of MLH1 staining were subjected to EpiTYPER for quantitative methylation of the MLH1 promoter region using the Agena MassARRAY® platform. Forty-one (28.28%) cases showed complete loss of tumour nuclei staining for ≥ 1 immunohistochemical stains. Twenty cases (13.79%) showed loss of MLH1/PMS2, 16 cases (11.03%) demonstrated isolated MLH1 loss and one case (0.69%) showed loss of MLH1, PMS2 and MSH6. Two cases showed isolated MSH6 loss and 2 cases showed loss of MSH2 and MSH6. Thirty-seven (90.24%) of 41 mismatch repair-deficient cases showed MLH1 loss. Most (83.78%) of these showed promoter hypermethylation, suggesting a sporadic occurrence of carcinogenesis. The patients from whom the two cases with isolated MSH6 loss, two cases with MSH2/MSH6 loss, and the six cases whereby MLH1 was not explained by methylation should be identified and offered genetic counselling with a view to possible germline mutation assessment as these patients are suspected of having Lynch syndrome. Thus, this study demonstrates a possible 10/145 (6.90%) patients who may have LS and require further testing and suggests a need to screen possible LS patients in the South African population.

Keywords::

• endometrial carcinomas
• Lynch syndrome
• South Africa

Introduction

Carcinomas of the endometrium are commonly encountered tumours in western countries.^1–4 The 2019 predicted incidence of endometrial carcinoma in the United States of America is approximately 7%.⁵ According to the latest published statistics for 2014 by the National Cancer Registry in South Africa, malignancies of the uterine corpus accounted for an approximate incidence of 3% amongst all tumours recorded in females.⁶

Microsatellite instability (MSI) is a frequent molecular anomaly in endometrioid endometrial carcinoma.^7–11 Most cases of microsatellite instability are due to alterations in expression of proteins of the DNA mismatch-repair system; namely MLH1, MSH2, MSH6 and PMS2.^11,12 It may be common practice to screen endometrial cancers for loss of these proteins to detect mutations associated with Lynch syndrome, or to determine if a patient is eligible for targeted immunotherapy with pembrolizumab, an anti-PD-1 agent, in some countries.^13–16 This is not the case in South Africa. The most common loss is seen in MLH1 as a consequence of sporadic promoter methylation, which, if detected, points towards a sporadic occurrence of the tumour and generally excludes Lynch syndrome.¹⁵ Patients in whom methylation is not detected may be referred for genetic counselling with a view to further mutational testing.¹⁵

The reported prevalence of Lynch syndrome varies by population and had been shown to range from 0% to 10%.¹⁷ For example, Finland has been identified as having well-documented mutations of MLH1.¹⁷ Studies by Hampel et al., however, have noted that the incidence of Lynch syndrome identified in endometrial carcinoma is comparable to colorectal carcinoma in patients from the same geographic locale.¹⁷ A study by Vergouwe et al. identified that over 20% of colorectal carcinomas in a region of South Africa demonstrated MLH1 or MSH2 loss and, of these, approximately 10.5% were calculated to be a result of germline mutations.¹⁸ Their study highlighted the possibility of a high incidence of inherited tumours within the South African population. It is thus possible that a greater incidence of germline mutation associated tumours occur in South African patients. Confirmatory germline mutational assessment is currently not available in public hospitals in the Gauteng province in South Africa. Thus, the cases in this study that have been identified as having mismatch repair deficiencies by immunohistochemistry cannot undergo confirmatory, definitive testing for Lynch syndrome. To the best of the authors’ knowledge, a study assessing for possible Lynch syndrome in endometrial carcinomas in the South African population has not been performed to date.

Materials and methods

Following ethical clearance (Clearance certificate number M151051), 145 cases of archived endometrioid endometrial carcinoma for the period 2009–2015 were retrieved from the Department of Anatomical Pathology, University of the Witwatersrand/National Health Laboratory Service at the Charlotte Maxeke Johannesburg Academic Hospital. This department receives cases from a large catchment area of public hospitals and clinics. The samples included biopsies and excision specimens. There were 57 excision specimens out of 145 and the remainder were curettage specimens. These specimens accounted for all cases of endometrioid endometrial carcinoma seen during the period 2009–2015. Our department drains all public hospitals and clinics in the Southern Gauteng area. As such, we receive all pathology specimens from public hospitals in this area. The cases were identified using a SNOMED search on the DISA and LabTrack laboratory information system.

Unfortunately, following curettage, many patients are lost to follow-up in the public hospital setting in South Africa. The cases were not submitted as consults from other pathology departments. The slides were reviewed, and the histological diagnosis confirmed. This study was confined to cases of endometrioid endometrial carcinoma as previous work has shown that microsatellite instability is in essence limited to this subtype of endometrial tumour.¹⁹ Age was categorised as < 60 and ≥ 60 years. In general, 50 years of age is generally accepted as the age of menopause; however, an age cut-off of 60 was used as a low percentage of patients (2/145) were under the age of 50.²⁰

Immunohistochemistry

The four mismatch repair antibodies, namely MLH1 (Novocastra, UK), PMS2 (Novocastra, UK), MSH2 (Novocastra, UK) and MSH6 (Novocastra, UK), were used to stain 4 μm deparaffinised sections. A microtome was used to cut tissue sections from paraffin-embedded blocks. The tissue sections were then floated onto slides, which subsequently underwent drying overnight at 60°C. These four immunohistochemical stains had previously been optimised according to departmental standard operating procedure. Staining of tissue sections includes the treatment of primary antibody serum. Each of the tissue sections then underwent staining with the following antibodies: MLH1 (Clone ES05, 1:50), PMS2 (Clone MOR4G, 1:50), MSH2 (Clone 25D12, 1:50) and MSH6 (PU29, 1:50). A mouse linker resulted in increased expression of antigens. The sections were washed with Tris buffered saline (TBS) at pH 7.6. Immunohistochemistry was undertaken using an automated staining machine (DAKO Autostainer Link 48, Denmark). A ready-made solution, the EnVision™ FLEX target Retrieval Solution, High pH, was used for antigen retrieval. The chromogen used was 3,3′ diaminobenzidine hydrochloride solution (DAB, Sigma, USA), which produced a brown pigment. Meyer's haematoxylin was the counterstain used on tissue sections. Appropriate positive and negative control tissue sections were used.

Immunohistochemically stained tumour tissue sections were then assessed in a binary manner. The tumour cells demonstrated either positive staining of their nuclei (irrespective of intensity or volume) or the tumour cells were negative as there was no staining of tumour cells, in the presence of an internal positive control, which took the form of staining of endothelial cells, lymphocytes and stromal cells. As the possibility existed of only a few foci of positive staining nuclei identified in certain cases, implying that the mismatch repair genes remained intact, tissue microarrays had not been performed.

Methylation assessment

MLH1 promoter hypermethylation analysis was performed using MassARRAY EpiTyper analysis, which was undertaken by Agena Bioscience at Inqaba Biotec.

EpiTYPER is a system using DNA methylation analysis technology, which has shown that it is a quantitative method of DNA methylation analysis that provides reliable results. The target sequence of the MLH1 promoter region, (−248 to −178), according to Pérez-Carbonell et al.,²¹ was entered into the software package, which then identified primers allowing for the best possible DNA coverage. The forward primers were: AGGAAGAGCGGATAGCGATTT and the reverse primers were: TCTTCGTCCCTCCCTAAAACG. A product yield of 187 base pairs was noted and allowed for evaluation of 11 CpG islands. There were 3 CpG targets that could not be assessed due to lower mass cleavage products, which is a known disadvantage of EpiTYPER.²² The remaining 8 CpG sites could be assessed. One (1) μg of genomic DNA was required for bisulphite treatment and a ratio of between 1.7 and 2.0 was recorded for each sample using a Nanodrop spectrophotometer. DNA tubes were stored at 4°C. The forward and reverse primers were reconstituted with nuclease-free water at equimolar concentrations of 100 μM. These were subsequently diluted to a working solution of 1 μM (personal communication, Inqaba Biotec).²³ Bisulphite conversion, PCR, in-vitro transcription and cleavage was performed according to the manufacturer's instructions.²³ The cleaved products were identified using the MassARRAY compact mass spectrometer and real-time software by Agena. Matrix-assisted laser desorption/ionisation time of flight (MALDI-TOF) mass spectrometry identified patterns that were created from different cleavage products of methylated or unmethylated target regions. The relative amount of methylation was calculated by comparing the signal intensity between mass signals from unmethylated and methylated template DNA. In each case, a percentage of methylation was calculated and depicted in an Epigram.

Results

Table 1 shows that the average age of patients diagnosed with endometrial carcinomas was 65.2 years, with the majority (66.9%) of patients being over the age of 60 years. Most of the endometrial carcinomas in the present study showed grade 2 Fédération Internationale de Gynécologie Obstétrique (FIGO) histological features.

Table 1: Age and pathological distribution of the study samples

Parameter/clinical characteristic	Number (%) (n = 145)
Age (years), mean ± SD	65.2 (± 9.2) years
< 60 years (reproductive age group)	48 (33.1)
≥ 60 years (post-menopausal age group)	97 (66.9)
Immunohistochemical loss of staining	41 (28.28)
Retained staining	104 (71.72)
Tumour grade		Mean age ± SD*
1	44 (30.34)	64.4 ± 7.8
2	63 (43.45)	65.5 ± 10.2
3	38 (26.21)	65.8 ± 9.2

*Analysis of variance p-value = 0.76.

Forty-one (28.28%) of the 145 cases demonstrated complete loss of staining of tumour nuclei for one or more immunohistochemical stains in the presence of adequate internal controls such as positive staining of stromal cells, endothelial cells and lymphocytes. There were 20 cases (13.79%) that showed loss of MLH1 and PMS2, whilst 16 cases (11.03%) showed isolated loss of MLH1 staining. A single case (0.69%) showed loss of three markers, namely MLH1, PMS2 and MSH6. There were two cases that demonstrated isolated loss of MSH6 staining and two cases showed loss of staining of both MSH2 and MSH6. Thus, MLH1 demonstrated loss of staining most commonly amongst the four antibodies. One hundred and four (71.72%) cases demonstrated retention of staining of all mismatch repair immunohistochemical markers. Figures 1–4 demonstrate tumours stained with the four immunohistochemical markers.

Figure 1: Lymphocytes and stromal cell nuclei stain positively (arrows) but the nuclei of tumour cells are negative. 4 μm MLH1 stained section. Original magnification: 200x.

Figure 2: Lymphocytes and stromal cell nuclei stain positively (arrows). Nuclei of tumour cells are negative. 4 μm PMS2 stained section. Original magnification: 200x.

Figure 3: There is positive staining of tumour nuclei. 4 μm MSH2 stained section. Original magnification: 200x.

Figure 4: Positive staining of tumour nuclei is demonstrated. 4 μm MSH6 stained section. Original magnification: 200x.

Table 2 demonstrates a statistically significant association between the tumour grade and the absence of IHC staining (p = 0.002). Of the 41 patients with an abnormal/loss of IHC staining, 22% had a grade 1 tumour, whilst 65.9% of the 41 cases with abnormal immunohistochemical staining showed grade 2 histological features. Approximately double the proportion of cases had abnormal immunohistochemical staining with grade 2 histological features than those with normal IHC staining. Of the 41 cases with abnormal IHC staining, 12.2% were grade 3 tumours.

Table 2: Comparison of age and tumour grades of immunohistochemically detected loss of staining and those with retained nuclear staining

Factor	Retained staining		Loss of staining		Total		Test statistics	p-value
	n = 104	%	n = 41	%	n = 145	%
Age (years) (mean ± SD)	65.2 ± 9.1		65.3 ± 9.7		65.0 ± 9.2		−0.0619^^	0.950
< 60	33	31.7	15	36.6	48	33.1	0.313^#	0.580
≥ 60	71	68.3	26	63.4	97	66.9
Tumour grade:
Grade 1	35	33.7	9	22.0	44	30.3	12.2^#	0.002
Grade 2	36	34.6	27	65.9	63	43.5
Grade 3	33	31.7	5	12.2	38	26.2

SD: standard deviation; ^# chi-square test; ^{^} Student's t-test.

The proportion of cases with retained IHC staining in FIGO grades 1 and 3 endometrial carcinomas was approximately four times more than those demonstrating loss of staining for their respective tumour grades. Grade 2 tumours showed a moderately increased proportion of IHC staining compared with tumours that had loss of IHC staining. One of the 37 cases that showed MLH1 loss that underwent hypermethylation analysis had insufficient DNA for further analysis. There were 31 (83.78%) cases that demonstrated methylation levels of more than 10% and there were 5 cases (13.51%) that had very low levels of methylation and were thus regarded as not being methylated.²⁴

Table 3 demonstrates that there were no statistically significant differences in the age of those who harboured methylated tumours versus those who did not (p = 0.532).

Table 3: Comparison of age and tumour grades of hypermethylated and unmethylated tumours

Factor	No methylation		Methylation		Not enough DNA		Total		p-value
	n = 5	13.51%	n = 31	83.78%	n = 1	2.70%	n = 37	%
Age (years)
< 60	1	20.00	10	32.26	1	8.33	12	32.43	0.532*
≥ 60	4	80.00	21	67.74	0	0.00	25	67.57
Tumour grade:
Grade 1	3	60.00	6	19.35			9	24.32	0.348*
Grade 2	2	40.00	22	70.97			24	64.86
Grade 3	0	0.00	3	9.68			3	8.11

*Fisher's exact test.

More than two-thirds of methylated cases demonstrated grade 2 histological features in contrast to unmethylated cases.

For comparison with the present study, only numbers of endometrioid endometrial carcinoma in other studies are shown as some of these studies examined all histological subtypes of endometrial carcinoma, whilst other studies focused only on endometrioid endometrial carcinomas (Table 4).^2,17,25–33 The present study shows that a high percentage of MMR-deficient cases of endometrioid endometrial carcinoma are due to MLH1 promoter methylation, similar to studies by other authors. The cases that are unmethylated raise concern for a possible underlying germline mutation, which in many of the western studies have been investigated.

Table 4: Comparison of some previously reported endometrial carcinomas in the literature with the present study

Author	Year	Sample size of endometrial carcinomas	Mean age of pts	Tumour FIGO grades	No. of patients with IHC MMR-deficient	Methylation of MLH1 promoter region
Current study	2019	145	65.2	1: 44	41/145 (28%)	31/36 (86%)
				2: 63
				3: 38
Bartley^Citation25	2011	45	55	Not stated for endometrioid endometrial carcinomas	Not stated for endometrioid endometrial carcinomas	Not stated for endometrioid endometrial carcinomas
Bruegl^Citation26	2014	46	57*	1 and 2: 35	46	35/46 (76%)
				3: 11
Bruegl^Citation27	2017	158	61.3*	1: 19	Not stated for endometrioid endometrial carcinomas	Not stated for endometrioid endometrial carcinomas
				2: 139
				3: 55
Cornel^Citation28	2018	51	63	1:39	Not done	19/51 (37%)
				2: 4
				3: 0
				Unknown: 8
Hashmi^Citation29	2019	126	55.1	1: 56	56/126 (44%)	Not tested
				2: 61
				3: 9
Hampel^Citation17	2006	421	60.9*	Not stated	85/127 (67%)	80/85 (94%)
Kim^Citation30	2018	173	54	1: 89	45/173 (26%)	18/26 (69%)
				2: 55
				3: 28
Nagle^Citation31	2018	598	59.5	1 and 2: 115	142/728 (20%)	109/728 (15%)
				3: 25
Peterson^Citation2	2012	61	65	1: 33	22/61 (36%)	33/61 (54%)
				2: 16
				3: 12
Stelloo^Citation32	2016	696	69*	Not stated for endometrioid endometrial carcinomas	196/696 (28%)	126/130 (97%) of microsatellite high case
Woo^Citation33	2014	77	58.6	1: 22	15/77 (20%)	Not tested
				2: 45
				3: 10

*Mean age of all histological subtypes of endometrial carcinoma.

Discussion

Hampel et al. have shown that sensitivity of family history is not a useful method of screening for possible patients with Lynch syndrome.³⁴ In a country such as South Africa, which is multiculturally rich, it is possible that language differences may be a limiting factor in acquiring an in-depth, extensive history of tumours in family members. This may then fare poorly as a screening utensil for Lynch syndrome. Ryan and co-workers as well as Garg and colleagues have noted that approximately 50% of germline mutation carriers with endometrial carcinoma do not demonstrate histological features that have been associated with inherited mutations.^35,36 The histological features include location of tumours in the lower uterine segment, tumour infiltrating lymphocytes, heterogeneity of tumour and peri-tumoural lymphocytes.^35,36 Shia has also stated that histological findings may not serve as a useful screening tool.³⁷ In the present study, the histological features could not be assessed in each case as not all patients underwent hysterectomy. This may be ascribed to many patients being lost to follow-up after their initial curettage. It is therefore not possible to discuss findings in the local population. However, by adopting an endometrial carcinoma screening programme, anatomical pathologists are placed in a position whereby they may be able to identify possible patients with Lynch syndrome in South Africa.

Methylation analysis by EpiTYPER demonstrated that the majority of cases (83.78%) of the abnormal/deficient MMR stains could be attributed to hypermethylation. This is consistent with findings of studies undertaken in western societies.^2,38 Identification of hypermethylation in these patients suggests a sporadic occurrence of carcinogenesis and, as such, these patients would not undergo additional molecular tests.

An article by Hashmi et al. has documented isolated MLH1 loss in a cohort of patients with colorectal carcinoma.³⁹ In their study, they noted that the cases demonstrating isolated MLH1 loss would undergo methylation analysis and BRAF mutational assessment. The identification of any degree of nuclear staining was interpreted as positive in the present study. Thus, even faint, focal staining of tumour nuclei was interpreted as MMR proficient. As such, this accounts for the identification of PMS2 positive staining despite loss of MLH1. This could possibly be attributed to missense mutations as noted by Shia, who stated that missense mutations may culminate in inactive mutant proteins that may still retain their antigenicity.⁴⁰ However, as mutational analysis had not been performed, it is not possible to state this with certainty. Furthermore, the clone of antibody used for PMS2 at the time that this study was performed is not a clone that has provided the best results as indicated by NordiQC⁴¹ and, as such, this may have resulted in an overly sensitive stain. Whilst the clones for the antibodies PMS2 (clone MOR4G), MSH2 (clone 25D12) and MSH6 (clone PU29) have not been assessed to be the best clones for their respective antibodies in assessments by NordiQC, these antibodies have been optimised for diagnostic use in our laboratory and have demonstrated adequate staining of internal controls such as endothelial cells, stromal cells and lymphocytes in control and test tissue samples.^41–43 Nevertheless, the results of the present study suggest that the possibility of PMS2 staining, although slight, may still exist in the absence of MLH1 staining.

Of the IHC diagnosed MMR-deficient tumours in the present study, the mean age at diagnosis was 65.3 (± 9.7) years (Table 2), which is higher than that noted in studies by Buchanan et al. and Kato et al., but is fairly similar to that seen by McConechy et al.^44–46 Of the IHC-deficient cases, grade 2 histological features were seen in 65.9% of samples in the present study, whilst the minority of tumours (12.2%) showed grade 3 features. Grade 2 histological features were also identified as being the most common subtype in a study by Bruegl et al. and McConechy et al.^{46, 47} This may be due to microsatellite unstable endometrial carcinomas being coupled to higher grade and stage.⁴⁸ In contrast, a study undertaken by Buchanan et al. demonstrated that grade 1 histological features were the most commonly occurring subtype.⁴⁴ The present study suggests a need to screen possible Lynch syndrome patients in the South African population. It is suggested that screening be performed in patients below 70 years of age as most patients in the current study were over 60 years of age and therefore, if screening is only for patients under the age of 60 years, many patients with possible Lynch syndrome may not be detected. In an attempt to curb the cost of tests, biopsies from such patients may undergo initial IHC testing for only MSH6 and PMS2, as suggested by Cho et al.¹⁵ Should abnormal/loss of staining in MSH6 and/or PMS2 IHC stains be documented, these may be followed by immunohistochemical testing of MSH2 and/or MLH1. Many pathology laboratories have access to immunohistochemical tests and the cost-effectiveness of IHC testing has been demonstrated. These factors together with a sensitivity and specificity of 83% and 89% respectively for Lynch syndrome are pointers to IHC being a useful screening tool.¹⁵ Subsequent reflex testing for promoter methylation may then be performed when indicated. Loss of staining of specific immunohistochemical markers may then facilitate testing of specific genes, should such tests be available.¹⁵ Identification of mismatch-proficient versus mismatch-deficient tumours may facilitate consideration of alternative treatment options as tumour cells of mismatch repair-proficient tumours are sensitive to methylating agents whereas mismatch repair-deficient tumours are not.⁴⁹ In addition, with the option of targeted therapy being available, the presence of mismatch-repair mutations allows for consideration of use of an antibody directed against programmed cell death (PD-1), which has shown favourable results.^11,50

Conclusion

In total, 41/145 cases demonstrated mismatch repair deficiency by immunohistochemistry, of which 37 (90.24%) showed MLH1 loss. Most (83.78%) MLH1 deficient cases demonstrated promoter hypermethylation, suggesting sporadic carcinogenesis. The two cases with isolated MSH6 loss, two cases with MSH2 and MSH6 loss, and six cases with deficient MLH1staining, unexplained by methylation, are suspected of having Lynch syndrome.⁵¹ Thus, a possible 10/145 (6.90%) patients may have LS requiring further testing. Two studies suggested a greater incidence of germline mutation associated tumours in South African patients compared with the Western world.^18,52 Our study aims to highlight the possibility of Lynch syndrome associated endometrial carcinomas and envisages that state patients with suspected germline tumours, requiring mutational assessment, may be offered tumour specific therapy.

Disclosure statement

No conflict of interest was reported by the author(s).

Additional information

Funding

This research has been funded by the University of the Witwatersrand FRC grants to individuals, University of the Witwatersrand SEED funding, NHLS Research Trust and NRF Thuthuka Grant.