MLH1-methylated endometrial cancer under 60 years of age as the "sentinel" cancer in female carriers of high-risk constitutional MLH1 epimutation

Abstract

Objective: Universal screening of endometrial carcinoma (EC) for mismatch repair deficiency (MMRd) and Lynch syndrome uses presence of MLH1 methylation to omit common sporadic cases from follow-up germline testing. However, this overlooks rare cases with high-risk constitutional MLH1 methylation (epimutation), a poorly-recognized mechanism that predisposes to Lynch-type cancers with MLH1 methylation. We aimed to determine the role and frequency of constitutional MLH1 methylation among EC cases with MMRd, MLH1-methylated tumors.

Methods: We screened blood for constitutional MLH1 methylation using pyrosequencing and real-time methylation-specific PCR in patients with MMRd, MLH1-methylated EC ascertained from (i) cancer clinics (n = 4, <60 years), and (ii) two population-based cohorts; "Columbus-area" (n = 68, all ages) and "Ohio Colorectal Cancer Prevention Initiative (OCCPI)" (n = 24, <60 years).

Results: Constitutional MLH1 methylation was identified in three out of four patients diagnosed between 36 and 59 years from cancer clinics. Two had mono-/hemiallelic epimutation (∼50% alleles methylated). One with multiple primaries had low-level mosaicism in normal tissues and somatic "second-hits" affecting the unmethylated allele in all tumors, demonstrating causation. In the population-based cohorts, all 68 cases from the Columbus-area cohort were negative and low-level mosaic constitutional MLH1 methylation was identified in one patient aged 36 years out of 24 from the OCCPI cohort, representing one of six (∼17%) patients <50 years and one of 45 patients (∼2%) <60 years in the combined cohorts. EC was the first/dual-first cancer in three patients with underlying constitutional MLH1 methylation.

Conclusions: A correct diagnosis at first presentation of cancer is important as it will significantly alter clinical management. Screening for constitutional MLH1 methylation is warranted in patients with early-onset EC or synchronous/metachronous tumors (any age) displaying MLH1 methylation.

Keywords: Constitutional MLH1 epimutation; Endometrial cancer; Lynch syndrome; MLH1 methylation.

Figure 1.. Schematic of overall study design and assays for constitutional MLH1 methylation analyses.

A: Flow diagram showing the ascertainment, selection, and numbers of patients eligible for testing for constitutional MLH1 epimutation. B: Map of the MLH1 CpG island proximal promoter region that corresponds with transcriptional activity,[6, 43] and assays designed to interrogate methylation status and levels in bisulfite-converted genomic DNA (not to scale). Purple asterix (*) indicates the quantitative CpG pyrosequencing and high-sensitivity real-time methylation-specific PCR (qMSP) assays used for first-pass screening to detect constitutional MLH1 methylation. Assays shown in gray are unbiased with respect to methylation status, such that PCR amplification will occur from both unmethylated and methylated templates. Assays in black are methylation-specific PCR-based (MSP) with primers overlapping designated CpG sites, hence amplification will occur only from methylated templates with high sensitivity. The MSP1 assay used for bisulfite-sequencing uses the same primers as the qMSP assay used for screening. MSP2 and MSP3 were used to confirm the presence of low-level methylation. Additional assays shown were used for confirmation of methylation. Horizontal bars show PCR products, circles show CpG sites interrogated within each amplicon. Orange line and squares indicate the location of the promoter c.−93G>A SNP (rs1800734) used to trace allele specificity for methylation in heterozygous cases positive for constitutional MLH1 methylation. The unbiased clonal bisulfite-sequencing assay was used to confirm and identify allelic methylation in germline DNA in patients with high levels of methylation measured by CpG pyrosequencing, and in tumor DNA. MSP3 was used to determine allelic methylation in heterozygous cases with low-level methylation. Locations of each assay are numbered with respect to the translation start site at +1 of MLH1 transcript, GenBank accession number NM_000249.3.

Figure 2.. Detection of monoallelic constitutional MLH1 epimutation in Proband 192.

A: Pedigree of Patient 192. EC, endometrial cancer; CRC, colorectal cancer; unk, unknown. For privacy, members of younger generations are not included (none have had a cancer diagnosis). B: Real-time methylation-specific PCR (qMSP) was performed from c.−188 to c.−403within the MLH1 promoter on bisulfite-converted genomic DNA from the nuclei of peripheral blood leukocytes (PBL) and saliva. Amplification curves show methylated MLH1 (red), which amplifies only when methylated DNA is present, run in parallel with MYOD (blue), which serves as a quality control measure for sample input and integrity. The percentage of methylated reference (PMR) value is shown. The high-resolution melt curve of the MLH1 amplicon for each sample indicates specificity for methylated amplicons with melt peak at 76±0.5 °C. C: Pyrosequencing traces are shown for five CpG sites from c.−241 to c.−272 of the MLH1 promoter in bisulfite-converted genomic DNA from peripheral blood leukocyte (PBL) nuclei and saliva. Methylation is detected by the presence of a peak at the cytosine (C) within each CpG site interrogated (gray bars), whereas unmethylated cytosines are detected as thymine (T) peaks within the same CpG sites, due to the conversion of unmethylated cytosines to uracils using bisulfite treatment. The assay measures the relative levels of methylated C against unmethylated cytosines at each CpG site interrogated, and reports these as a percentage of methylation value above. The mean level of methylation across all five CpG sites is calculated and shown above. The yellow bar indicates a non-CpG cytosine used as a quality control measure to ensure complete bisulfite-conversion to T, whereupon this yields a valid test result. Methylation was measured at 48.2% in PBL and 47.4% in saliva. D: Left, Sanger sequencing electropherogram showing Patient 192 was heterozygous for the MLH1 promoter c.−93G>A SNP (rs1800734). Right, pictogram of clonal bisulfite sequencing across a fragment of the MLH1 CpG island promoter region from c.−48 to c. −370 showing the methylation status at 16 individual CpG sites (circles) within individual alleles (horizontal lines). The rs1800734 SNP genotypes are represented by colored squares on each allele. This SNP is flanked by CpG dinucleotides numbered correspondingly in the electropherogram (left) and pictogram (right). Hypermethylation was restricted to alleles bearing the ‘A’ genotype at rs1800734, indicating monoallelic methylation.

Figure 3.. Detection of hemiallelic constitutional MLH1 epimutation in EC Proband 213.

A: Nuclear pedigree of patient 213, showing a non-Lynch syndrome cancer history on the paternal side. B: Left, illustrative amplification curves showing positive amplification of methylated MLH1 by real-time methylation-specific PCR (qMSP) within the MLH1 promoter on bisulfite-converted genomic DNA. Right, melt curve of the amplicons, confirms product specificity at a melt temperature of 76±0.5 °C. C: CpG pyrosequencing traces with legend according to Figure 2. MLH1 methylation measured between 46.8% to 52.4%. D: Clonal bisulfite sequencing of a larger fragment of the MLH1 promoter confirms the presence of methylation in a hemiallelic pattern, consistent with the CpG pyrosequencing result. The patient was homozygous “G” at the rs1800734 SNP, so the parental allele-of-origin of the constitutional MLH1 methylation could not be determined.

Figure 4.. Detection of low-level constitutional MLH1 methylation mosaicism in Patient 166.

A: CpG pyrosequencing yields negative test results in genomic DNA from peripheral blood leukocytes (PBL) nuclei and saliva, but low levels of methylation are detectable in DNA extracted from other sources of histologically normal tissue samples (macrodissected formalin-fixed paraffin-embedded) from surgically resected organs, as labeled. B: Real-time methylation-specific PCR (qMSP) shows positive amplification of methylated MLH1 templates in PBL and saliva samples with methylation levels too low to be detectable by CpG pyrosequencing, but absence of methylation in hair follicles. C: Top, partial sequence within the MLH1 promoter shows heterozygosity for the c.−93G>A SNP (arrow). Dashed lines show the locations of individual CpG sites flanking the SNP. Beneath, Sanger sequencing of methylation-specific PCR (MSP) products across the same region encompassing the c.−93G>A SNP site after bisulfite-conversion of genomic DNA shows only methylated templates were amplified (as C is retained at CpG sites, whereas isolated Cs at non-CpG sites are converted to T [asterix]) and these are exclusively linked to the G allele at c.−93 SNP position. This indicates the low-level methylation (amplifiable by MSP) is restricted to the G allele.

Figure 5.. Low-level mosaic constitutional MLH1 methylation predisposes to multiple MLH1-methylated primary tumors in Patient 166.

A: CpG pyrosequencing confirms the presence of significant levels of MLH1 methylation in all three primary tumors, including the small intestine adenocarcinoma, which had not previously been assessed for this. B: Clonal bisulfite sequencing within the MLH1 promoter encompassing the c.−93G>A SNP (rs1800734) using primers unbiased with respect to methylation status shows the patterns of methylation in the tumors (left). The skin sebaceous carcinoma and small intestine tumors had methylation on both alleles, although few A alleles at c. −93 remained in the skin tumor. The EC was monoallelically methylated on the G allele. The same assay performed on accompanying normal tissue samples (right) only detected a single methylated molecule in endometrial epithelium, which was the G allele at c. −93G>A. C: The c. −93G>A SNP was used to trace allelic representation in each primary cancer (right) with reference to a paired normal tissue sample (right). Partial Sanger sequences are shown across the c. −93G>A SNP, with measurements of each allele in relative fluorescence units (RFU) taken from the peak for each genotype shown below. Loss-of-heterozygosity (LOH) was assessed by calculating (A/G_Tumor)/(A/G_Normal). If A/G was ≤0.6 in the tumor, this was designated as LOH of the A allele. Retention of heterozygosity (ROH) was designated if A/G >0.6 in the tumor.

Figure 6.. Detection of low-level constitutional MLH1 methylation in OCCPI patient EC-32.

A: CpG pyrosequencing traces are shown DNA from whole blood, uninvolved endometrial tissue, and tumor, as labeled. The mean methylation value is shown above. Methylation was undetectable in whole blood DNA, given this was below the limit of detection (2.3%), but was detectable at low levels (2.8%) in uninvolved normal endometrial tissue and at high levels (65.6%) in the tumor. B. Real-time methylation-specific PCR (qMSP) within the MLH1 promoter on bisulfite-converted genomic DNA from whole blood (left) and confirmed in uninvolved normal endometrial tissue (right). Amplification curves show methylated MLH1 (red), which amplifies only when methylated genomic DNA template is present, run in parallel with MYOD (blue), which serves as a quality control measure for sample input and integrity. An exemplary high resolution melt curve of the MLH1 amplicon is shown for the whole blood sample (middle), indicating specificity for methylated amplicons with melt peak at 76±0.5 °C. C: Electrophoresis gels of traditional MLH1 methylation-specific PCR (MSP) amplification products according to Figure 1. Left, MSP1 used the same primers as the qMSP. Right, MSP2 was conducted in a distinct region of the MLH1 CpG island. Both MSP reactions included DNA from whole blood (WB) of EC-32, a healthy control (HC, negative control), RKO colorectal cancer cell line (positive control), and a sample with no template (NT) added. Beneath, pictograms of clonal bisulfite sequencing of each MSP amplicon to determine the methylation status at 16 individual CpG sites within the respective regions. Black circles show methylated CpG sites and white circles show unmethylated CpG sites on individual molecules (horizontal line).