Mosaicism in Patients With Colorectal Cancer or Polyposis Syndromes: a Systematic Review

Abstract

Background & Aims:

Somatic mosaicism arises from genetic alterations that occur after the first division of the zygote, so that not in all cells in the body contain the same genetic variants. These variants might contribute to colorectal cancer (CRC) and polyposis syndromes. We performed a systematic review to provide a comprehensive overview of somatic mosaicism in patients with CRC and polyposis syndromes.

Methods:

We searched PubMed through March 2019 to identify reports of mosaicism in patients with CRC or polyposis syndromes. We divided the final set of studies into 3 subgroups describing APC mosaicism, mosaicism in other genes associated with susceptibility to CRC susceptibility, and epigenetic mosaicism.

Results:

Of the 232 articles identified in our systematic search, 46 met the criteria for further analysis. Of these, 35 studies described mosaic variants or epimutations in patients with CRC or polyposis syndromes. Nineteen studies described APC mosaicism, comprising a total of 57 patients. Six described mosaicism in genes associated with familial CRC syndromes, such as Lynch and Cowden syndromes. Ten studies described epigenetic mosaicism, sometimes resulting from a germline variant (such as deletion of EPCAM).

Conclusions:

In a systematic review, we found that many patients with polyposis syndromes have genetic mosaicism, most frequently in APC variants; this information can be used in management of patients. Mosaicism in genes associated with susceptibility to CRC contributes to development of other familial CRC syndromes. Heritable epigenetic mosaicism is likely underestimated and could have a dominant pattern of inheritance. However, the inheritance of primary mosaic epimutations, without an genetic cause, is complex and not fully understood.

INTRODUCTION

Somatic mosaicism occurs when a postzygotic genetic variant is present in only a proportion of cells in the body.^{1, 2} Mosaic variants have been shown to cause familial diseases, most commonly in disorders caused by genes with high mutational frequencies.³ These were first described in patients with Duchenne muscular dystrophy but have since been described in patients with several other diseases, including von Hippel-Lindau disease, neurofibromatosis type 1, and tuberous sclerosis.^4–8 Although mosaicism is an accepted phenomenon in these diseases, it is still under-recognized in the clinic, and guidelines for screening patients likely to carry mosaic variants are lacking.

The time at which variants arise during development determines their distribution and phenotypic effects.^{2, 9} For example, if a variant arises early during embryogenesis, it can be present in multiple germ layers and organ systems. Furthermore, whereas mosaic variants often result in a milder phenotypes in probands, they may be inherited by offspring in a heterozygous form, often leading to more severe phenotypes.¹⁰ Therefore, it is critical to identify mosaicism not only for the patients but for their offspring.

In this systematic review, we aim to provide a comprehensive overview of somatic mosaicism in patients with colorectal cancer (CRC) and polyposis syndromes and to develop a decision tree to guide clinicians in the management and genetic screening of patients likely to carry mosaic variants.

METHODS

A systematic bibliographic search was conducted to identify studies describing mosaicism in CRC and polyposis patients (Figure 1), in accordance with PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) 2009 guidelines.¹¹ First, an electronic search was performed in March, 2019 in PubMed using (mosaic OR mosaicism) AND (colon OR colorectal OR polyposis OR “adenomatous polyps” OR “colonic adenomas”). All output was manually inspected and reference lists from the selected articles reviewed. Inclusion criteria used were (1) positive or negative testing for a mosaic variant or epimutation and (2) mosaicism resulting in CRC or polyposis. Review articles, commentaries, and author replies were excluded from the analysis, but their reference lists were inspected. Additionally, a separate PubMed search using (PTEN OR STK11 OR SMAD4 OR TP53) AND (mosaic OR mosaicism) was performed to identify publications about mosaic variant carriers with gene variants in CRC susceptibility genes, irrespective of CRC status.

Figure 1:

Flowchart of article selection.

RESULTS AND DISCUSSION

Of the 232 records identified in the initial search, 187 were excluded. The majority of excluded papers were eliminated because they did not refer to mosaic variants or mosaic methylation but to the MOSAIC (Multicenter International Study of Oxoplatin/5-Fluorouracil/Leucovorin in the Adjuvant Treatment of Colon Cancer) trial or to diseases with mosaic in the name (e.g., mosaic variegated aneuploidy syndrome, caused by pathogenic variants in the BUB1 gene). One additional article describing mosaicism, which had not been found in the initial search, was identified through the manual inspection of the references of all reviews. Of the remaining 46 articles, six did not report patients with mosaicism but excluded mosaicism (n=2), described a heterozygous patient (n=1), or no genetic testing was performed (n=3).^12–17 In addition, two articles were methodic papers testing patients previously described,^{18, 19} and three articles reported multiple cell populations within a tumor (tumor heterogeneity), not mosaicism.^20–22 The final 35 records reporting mosaicism included 19 studies of APC gene mosaicism in patients with polyposis, six studies describing mosaic variants in other known CRC susceptibility genes, and ten studies describing epigenetic mosaicism. These three subgroups are discussed below.

APC mosaicism

Familial adenomatous polyposis (FAP) is a rare autosomal dominant disease that accounts for ~1% of CRCs.^23–25 FAP is caused by germline heterozygous variants in the APC gene and is associated with the development of hundreds to thousands of colorectal adenomas at an early age.^23–25 FAP can present in the classic, more severe form or as attenuated FAP (AFAP), a milder form characterized by fewer adenomas (20–100) and a later age of onset.^26–28 This attenuated phenotype may be caused by germline variants in the 5’ or 3’end of the APC gene or by mosaic APC variants that arise during embryogenesis (i.e., not inherited from the parents).^{10, 28–30}

In 1999, Farrington and Dunlop tested the DNA of five de novo FAP patients, as well as of their parents, and were the first to identify somatic APC mosaicism.³¹ One parent of one proband carried the pathogenic APC variant that was also present in the proband, albeit in mosaic form. Additionally, one of the five patients was found to carry the germline variant in only 22% of leukocytes and was thus considered mosaic.³¹ Since this first study, the phenomenon of APC mosaicism has become a widely accepted cause of AFAP and FAP. In 2007, it was estimated that de novo FAP is caused by somatic APC mosaicism in approximately 20% of patients.^{10, 30, 32} Given that one in every 10,000 people is estimated to have FAP, including 20–30% caused by de novo APC variants, the number of mosaic carriers is expected to be very high.^33–35 However, until now, only 19 papers have been published describing APC mosaicism, and nearly half of these papers (n=8) are case reports describing only a few patients (Table 1). In total, these studies describe 57 patients carrying mosaic defects (Supplemental Table 1).

Table 1:

Studies describing patients with mosaic APC variants

ID	Study	Type of patients tested	Mosaic patients/all patients tested	Type of mosaicism (# of patients affected)	Method	Patient phenotypes (# of patients affected)
1	Mandl et al, Hum Mol Genet. 199483	Family of FAP patient	1/1	A(1)	HD	FAP1
2	Farrington et al, Am J Hum Genet. 199931	Parents of de novo probands	2/5^	A(2)	Sanger, PCR cloning, single cell analysis	FAP (1), no adenomas (1)
3	Davidson et al, Hum Mutat. 200236	Parents of APC mutation carriers	1/1	D(1)	PCR, SSCP, sequencing, linkage analysis	FAP1
4	Aretz et al, Hum Mutat. 200730	De novo APC mutation carriers	8/75	A(8)	PTT, DHPLC, SNaPshot	AFAP (5), FAP 100–200 (3)
5	Hes et al, Gut. 200810	APC mutation carriers	10/242	A(8), B*(1), D(1)	DGGE/PTT	No polyps (1), AFAP(5), FAP 100s (4)
6	Schwab et al, Fam Cancer. 200837	Parents of de novo probands	1/1	D(1)	PCR, sequencing	No polyps
7	Kanter-Smoler et al, BMC Med. 200838	APC mutation negative de novo patients	1/3	A(1)	SSCP/HD, PTT	AFAP
8	Kaufmann et al, J Mol Diagn. 200984	Parents of proband	1/1	D(1)	Linkage (mutation paternal allele)	AFAP
9	Filipe et al, Dis Colon Rectum. 200985	AFAP patient	1/1	A(1)	PTT/sequencing	AFAP
10	Baert-Desumont et al, Am J Med Genet A. 201186	Parents of de novo proband	1/1	A(1)	Sequencing/ haplotype analysis	FAP 100–200
11	Necker et al, J Med Genet. 201187	Unexplained FAP patients	2/2	A(2)	PTT	FAP 100s (2)
12	Mongin et al, Clin Genet. 201288	De novo FAP patients without family history	1/17	A(1)	HRM, sequencing	AFAP
13	Iwaizumi et al, Hum Genome Var. 201589	Unexplained AFAP patients	1/1	A(1)	Sanger, NGS	AFAP
14	Yamaguchi et al, J Hum Genet. 201590	Unexplained AFAP patients	1/1	A(1)	WES, NGS	AFAP
15	Out et al, Fam Cancer. 201591	Unexplained AFAP patients (>10 adenomas)	2/161 + 2/2	A(2), B (1), B*(1)	HRM on leukocyte (161) HRM on tumor DNA (2)	AFAP (4),
16	Spier et al, J Med Genet. 201643	Unexplained (A)FAP patients (>20 synchronous adenomas/>40 non-synchronous adenomas)	5/20 + 2/80	A(7), B or C(1)	NGS on adenomas (20) WES on leukocytes (80)	AFAP (5), FAP 100–500 (2)
17	Jansen et al, Gastroenterology, 201714	Unexplained AFAP patients (20–100 adenomas)	9/18	A(3), B(5), C(1)	NGS on adenomas	AFAP (8)
18	Stormorken et al, Fam Cancer. 201840	AFAP patient	1/1	A(1)	Sanger, MLPA, NGS	AFAP
19	Ciavarella et al, Eur J Hum Genet. 201841	Unexplained (A)FAP patients >20 adenomas by age 35/>50 adenomas by age 55	4/8	B(3), B*(1)	Digital PCR, adenoma-first, WES on leukocytes	AFAP (2), FAP 100–200 (2)

Patients with familial adenomatous polyposis (FAP, >100 adenomas) and attenuated FAP (AFAP, 10–100 adenomas) were tested. Types of mosaicism included variants detected in leukocytes (A), variants present only in adenomas and not leukocyte (B), variants present in adenomas, leukocyte not tested (B*),variants present only in adenomas and not in normal colonic tissue (C), or mosaicism is hypothesized, not proven (D). The methods used were Sanger sequencing (Sanger), single-strand conformational polymorphism (SSCP), heteroduplex analysis (HD), denaturing gradient gel electrophoresis (DGGE), protein truncation test (PTT), denaturing high-pressure liquid chromatography (DHPLC), high-resolution melting analysis (HRM), multiplex ligation-dependent probe amplification (MLPA), next-generation sequencing (NGS), and whole exome sequencing (WES).

^{^}Mosaicism detected in one proband and one parent.

¹Patient also carried a second heterozygous germline APC variant. For patients with FAP, the number of polyps are indicated.

Initially, APC mosaicism was detected mainly by testing leukocyte DNA from the family members of probands with FAP.^{31, 36, 37} If multiple people in the same generation carried a de novo APC variant, mosaicism was hypothesized, but not necessarily proven, to be present in one of the parents.^{36, 37} Newer methods, such as the protein truncation test (PPT) and denaturing high-performance liquid chromatography (DHPLC), have enabled the detection of mosaic variants with small allele frequencies.^{10, 30, 38} Advances in DNA analysis, such as next-generation and even whole-exome sequencing methods, have enabled the detection of variants with frequencies <1% but have increased the likelihood of finding false positives.³⁰ A more specific method proposed by Aretz et al., in which multiple adenomas are tested for an identical variant, allows not only detection of allelic variants present at low frequencies but also of mosaicism confined to the colon.³⁰ Three studies using this “adenoma first” approach found mosaic variants in 14 patients with previously unexplained FAP, indicating that the incidence of mosaicism is likely underestimated.^39–41 The majority of patients with FAP currently reported to have mosaic APC defects (n=44, 77%) have a pathogenic variant present in DNA from leukocytes and adenomas (Table 1, Supplemental Table 1). These numbers are due, in part, to conventional screening methods, in which leukocyte DNA is tested for variants and mosaicism confined to the colon is undetected.

Three germ layers arise during development: mesoderm, endoderm, and ectoderm (Figure 2A).^{9, 42} Identifying the germ layer(s) in which a specific allelic variant is present helps determine the time during embryogenic development at which the variant arose. The endodermal and ectodermal layers, which go on to become colon tissue and buccal mucosa, respectively, differentiate by day 8 of embryonic development. Primordial germ cells are thought to arise from the ectodermal layer during the second week of development.^{9, 42} Therefore, variants present in both layers are presumed to have arisen before day 8 and before the primordial germ cells develop, and are more likely to be inherited by offspring.^{9, 42} The mesodermal layer, from which blood cells develop, becomes distinct later during embryogenesis, around day 16.^{9, 42} Therefore, variants present in both leukocytes (mesoderm) and colon tissue (endoderm) are presumed to have arisen before day 16 and thus may also be present in germ cells. In contrast, variants confined to the colon are thought to have arisen after day 16 and are unlikely to be inherited by offspring (Figure 2A).^{39, 41, 43} As shown in Figure 2A, determining which tissues carry the mosaic variant can give an indication of the moment the mosaic variant arose, and subsequently whether the variant could be inherited by offspring.

Figure 2:

A: Timing of the mosaic variant, adapted from Tuohy et al.³ B: Decision tree for the clinical management of patients with apparent de novo polyposis. BM = buccal mucosa, NM = normal colonic mucosa. Adapted from Hes et al¹⁰, key additions shown in grey.

As expected, the majority of patients with somatic mosaic APC variants (n=37, 65%) displayed an AFAP phenotype with 20–100 adenomas. However, 17 patients (30%) showed a FAP phenotype, and three patients (5%) had no adenomas, indicating a broad range of phenotypes in patients with APC mosaicism.

The presence of somatic APC mosaicism resulting in FAP or AFAP has consequences for clinical practice and genetic counseling. Although Hes et al. have already created a decision tree for the clinical management of patients with apparent sporadic FAP, new advances indicate that many patients with mosaicism are missed by screening only leukocyte DNA.¹⁰ We have generated an adjusted decision tree (Figure 2B), which takes all current APC mosaicism studies into account. In this adjusted decision tree, we advise to test adenomas from patients without an APC germline variant to detect a possible underlying mosaicism. Testing adenoma DNA will allow the detection of germline mosaicism of variants with very low allele frequencies, as well as mosaicism confined to the colon. Because inheritance of a mosaic variant would lead to FAP in the offspring, it is imperative to detect mosaic variants early.

In terms of clinical management, when a heterozygous mutation is found, patients should receive screening according to FAP guidelines, with colonoscopies every 1–2 years starting age 10–15.^{35, 44, 45} Colectomy should be considered when the polyp burden is too great.^{35, 44, 45} However, if the heterozygous mutation is found in the 5’ or 3’ part of the APC gene, patients will present with AFAP and should be screened accordingly with annual or biennial examinations starting from age 18–20 years.^{35, 44, 45} Mosaic carriers should be screened according to their phenotypes. For example, although the majority of mosaic carriers present with AFAP, if a FAP phenotype is observed, patients should be screened according to FAP guidelines.

Mosaic variants of other CRC susceptibility genes

Mosaic variants of other CRC susceptibility genes, such as MLH1, MSH2, PTEN, and STK11, have been described, although rarely (Table 2). Pathogenic germline variants of the mismatch repair (MMR) genes MLH1, MSH2, MSH6, and PMS2 result in Lynch Syndrome (LS), a dominant disorder characterized by MMR deficiency and microsatellite instability.^{46, 47} Only two LS patients with mosaicism have been described. In a study by Pastrello et al., one patient was found to carry a pathogenic variant present in the family, albeit with an allele frequency of only 41–47%. The authors hypothesized that this was not de novo mosaicism but “revertant mosaicism,” in which the inherited disease-causing mutation that was corrected by a genetic event that completely or partially restored the function of the gene.⁴⁸ However, the phenotype of the patient was not milder than that normally observed in patients with LS, so the revertant mechanism would likely have been missed and the mosaic patient would have been classified as having LS caused by a germline mutation. The second described mosaic LS patient carried an MSH2 variant detectable only by high-resolution melting (HRM) analysis, not by sequencing. The patient’s affected son (CRC54) carried the variant in his germline and presented with an LS phenotype. The family history was indicative of a mosaicism, as CRC was not observed in relatives other than the proband and her son. Although additional studies have hypothesized that more de novo LS patients could be due to mosaicism, to date, none have been described. Recent studies show that many patients have biallelic somatic inactivation of the MMR genes, which cause LS-like tumors with an MMR-deficient phenotype; it is possible that a percentage of these patients carry mosaic variants.^49–51 One recent study tested for mosaicism in a patient with identical MLH1 variants in two tumors, but mosaicism was excluded.¹⁴ Although the incidence of mosaic MMR variants might be low, with current screening guidelines carriers of mosaic variants may go undetected. To detect these mosaic MMR variants, we created a decision tree which advises testing the tumor for somatic MMR variants if neither a germline variant nor MLH1 promoter hypermethylation is found (Figure 3). Somatic variants in the tumor could indicate either biallelic somatic inactivation or an underling mosaic variant. To determine whether a variant present in the tumor is somatic or due to a mosaicism, additional tissue, such as buccal mucosa, normal colonic mucosa, or a second tumor if present should be tested.

Table 2:

Studies describing patients with mosaic variants in other CRC and polyposis syndromes

ID	Study	Syndrome (gene)	Patients tested	Mosaic patients/all patients tested	Method	Patient phenotypes	Additional notes
1	Sourrouille et al, Fam Cancer, 201392	LS (MSH2)	Mutation-negative suspected LS patients	1/18	Sanger, MLPA, HRM	CRC79 (MSH2-, MSI)	Variant passed to offspring (son with CRC54)
2	Pastrello, Am J Med Genet A, 200948	LS (MLH1)	Case report	1/1	Sanger, DHPLC	EC48	Revertant mosaicism
3	Salo-Mullen, Fam Cancer, 201481	CS/PHTS (PTEN)	Expected CS patient	1/1	aCGH, MLPA	Breast DCIS age 40, macrocephaly, age 43: hamartomatous and ganglioneuromatous intestinal polyps, lingual and labial papillomatosis, acral keratoses, uterine fibroids (age 28), fibrocystic breasts (age 36), mucosal fibromas, and visceral and cutaneous hemangiomas.	Variant frequency was intermediate between the normal control and the heterozygous control
4	Gammon, Clin Genet, 201356	CS/PHTS (PTEN)	Parent of proband	1/1	Sanger	Macrocephaly and Hashimoto’s thyroiditis	Variant present in <10% of leukocyte DNA reads. Daughter inherited variant and has full CS phenotype.
5	Pritchard, Genetics in Medicine, 2013	CS/PHTS (PTEN)	Expected CS patient	1/1	NGS	Lhermitte-Duclos disease, macrocephaly, mucosal papillomas, acral keratoses, hamartomatous polyps, ganglioneuroma, lipomas, thyroid goiter	Variant present in DNA of 1.7% of leukocytes; 50% of skin fibroblasts and tumor tissue; and 25% of colonic mucosa.
6	McKay, Fam Cancer, 201693	PJS (STK11)	Expected PJS patients	4/300	Sanger, MLPA	1: PJS age 37 (lip pigmentation, small and large bowel polyps), 2–4: clinical diagnosis of PJS	All patients had mosaicism in leukocytes, one patient passed the variant to the offspring.

Methods used were multiplex ligation-dependent amplification (MLPA), high-resolution melting (HRM), denaturing high-performance liquid chromatography (DHPLC), next-generation sequencing (NGS), array comparative genomic hybridization (aCGH), and Sanger sequencing (Sanger). Patient phenotype shows tumortype followed by age of onset. Patients presented with colorectal cancer (CRC), endometrial cancer (EC), and ductal cancer in situ (DCIS). Variants in these genes are linked to Lynch Syndrome (LS), Cowden Syndome (CS), PTEN harmatoma tumor syndrome (PHTS) and Peutz-Jeghers Syndrome (PJS).

Figure 3:

Decision tree for the clinical management of patients with a Lynch-like tumor MSI-H = high microsatellite instability, MMR = mismatch repair, BM = buccal mucosa, NM = normal colonic mucosa.* MLH1 promoter hypermethylation is sometimes tested before screening for germline variants.

Somatic mosaicism of PTEN variants has been associated with Cowden syndrome (CS). CS patients have a very broad range of phenotypes, including macrocephaly, mucocutaneous lesions, acral keratosis, papillomas, fibromas, and increased risk of breast, endometrial, thyroid, colon, skin, and renal cancers.^52–55 Of three patients with mosaic PTEN variants, only one had a clearly milder CS phenotype, restricted to macrocephaly and Hashimoto’s thyroiditis, whereas the proband’s heterozygous daughter (age 20) had more pronounced CS characteristics, such as macrocephaly, multiple ganglioneuromas, and multinodular goiter.⁵⁶ However, CS phenotypes are variable, so determining the extent to which the level of mosaicism correlates with phenotype severity can be difficult. Interestingly, two patients have been described as having both a germline PTEN variant and a mosaic PTEN variant, leading to a more severe Proteus-like Syndrome phenotype with lymphatic vascular malformation, macrocephaly, segmental overgrowth, epidermoid nevi, and fibrocystic breast cancer (Supplemental Table 2).^{57, 58}

One large study tested 300 patients suspected to have Peutz-Jeghers syndrome (PJS) and detected four patients with somatic mosaicism of STK11 variants. PJS, a familial syndrome characterized by the presence of hamartomatous polyps in the gastrointestinal tract and typical dark blue to dark brown pigmentation on the lips, buccal mucosa, hands, and feet, is typically caused by germline variants in STK11.^59–61 However, although all four patients received a clinical diagnosis of PJS, molecular diagnostics failed to detect germline STK11 variants. The study also found that a daughter of one of the mosaic patients inherited the STK11 variant, which was therefore heterozygously present in her germline DNA. Although phenotypic clinical data are lacking, it was noted that she showed phenotypic manifestations at age 4, as opposed to age 37, when her parent was diagnosed. This observation indicates that patients with STK11 mosaicism may exhibit milder phenotypes than patients with germline STK11 variants, but this needs to be confirmed in more patients.

Epigenetic mosaicism

Epimutations, although mitotically heritable, are often meiotically reversible and often show somatic mosaicism, with differing distributions of the altered epigenotype in various tissues.⁶² Constitutional epimutations, including mosaic epimutations, are described as primary (i.e., without an underlying genetic cause) or secondary (i.e., due to a genetic variant).⁶²

Perhaps the best known examples of heritable epigenetic mosaicism are due to EPCAM deletions, which result in mosaic MSH2 methylation. Studies of nearly 50 families (Table 3) have described more than 200 patients who carry germline 3’-end EPCAM deletions.^63–66 One of these studies estimated that ~20–25% of LS patients with cancers with immunohistochemical loss of MSH2 and MSH6 expression, without a germline MSH2 mutation (or 2–3% of all LS patients) carry a deletion in EPCAM.⁶⁷ However, the total number of EPCAM deletion carriers is likely higher than these studies indicate due to bias in our literature search, which included only studies that described EPCAM mutation status with respect to MSH2 methylation and excluded studies that purely reported EPCAM deletions, without studying the mosaic state of MSH2 methylation.^{67, 68} In EPCAM mutation carriers, the mosaic state of epigenetic MSH2 inactivation depends on the expression of the EPCAM gene, which is not equal in all tissues. For example, the expression of EPCAM is high in epithelial tissues⁶⁹, resulting in the methylation of MSH2 in tissues conventionally associated with LS.⁶⁴ This explains why families with EPCAM deletions display classical LS features even though MSH2 is not inactivated in all cells. Due to its underlying genetic nature, mosaic MSH2 methylation shows a dominant pattern of inheritance and imparts a CRC risk comparable to that of direct MSH2 mutations.^{64, 70}

Table 3:

Studies describing patients with mosaic epimutations in other CRC and polyposis syndromes

	Study	Type of patients tested	Mosaic/pat tested	Gene	Method	Phenotype patients	Primary/Secondary epimutation
1	Suter et al, Nature Genetics, 200476	Mutation negative polyposis or CRC patients	2/84	MLH1	COBRA	CRC<50 (2)	Assumed primary
2	Chan et al, Nature genetics, 200663	Early-onset or familial MSI CRCs	1/31 + 9 relatives	MSH2	MSP, haplotype analysis, pyrosequencing	CRC <50 (2) EC <50 (1) Adenomas>70 (1) No phenotype (6)	Secondary (EPCAM mutation carriers)
3	Morak et al, EJHG, 200894	Patients with MLH1-deficient MSI-H tumors	6/79	MLH1	MSP, MS-MPLA, SNP typing, haplotype analysis	CRC<50 (6)	Assumed primary (all 6), but 3/6 patients carried MLH1 c.−93G>A
4	Ligtenberg et al, Nature Genetics, 200864	LS-suspected families	7/unknown	MSH2	MLPA, SNP genotyping, MSP, pyrosequencing	CRC<50 (3) CRC50–55 (4)	Secondary (EPCAM mutation carriers)
5	Hitchins et al, Clin Genet, 201195	Mutation negative sLS patients	1/222	MLH1	qMSP, COBRA	CRC<50	Assumed primary, but MLH1 c.−93G>A present
6	Kempers et al, Lancet Oncol. 201165	Patients found through diagnostic screening	41 families, 194 mutation carriers^	MSH2	ND	CRC average age 43 (93)	Secondary (EPCAM mutation carriers)
7	Ward et al, Genet Med. 201396	Mutation negative sLS patients	6/416	MLH1	qMSP, sequencing	CRC<50 (3) CRC50–70 (3)	Primary (5), Secondary (1): (c.−27C>A variant)
8	Mur et al, Clin Genet, 201466	LS-suspected families	7 families, 28 mutation carriers. Methylation tested in 10/28.	MSH2	MLPA, MS-MLPA	CRC <50 (7), CRC52 (1), DC52 (1) Hg.ad28 (1)	Secondary (EPCAM mutation carriers)
9	Sloane et al, JAMA Oncol, 201580	Parent of proband	1/1	MLH1	Haplotype analysis, qMSP, pyrosequencing	Not affected	Assumed primary
10	Pinto et al, Cancer Med. 201897	Mutation negative polyposis or CRC patients	4/38	MLH1	MS-MLPA, qMSP, ddPCR, sequencing	CRC <50 (4)	Assumed primary

Methods used to determine methylation were combined bisulfite restriction analysis (COBRA), methylation-specific PCR (MSP), quantitative MSP (qMSP), single nucleotide polymorphism (SNP) genotyping, methylation-specific multiplex ligation-dependent amplification (MS-MLPA), digital droplet PCR (ddPCR), or not described (ND). Phenotype of patients is shown as the tumor type followed by the age of onset. Patients presented with colorectal cancer (CRC), endometrial cancer (EC), duodenal cancer (DC), and high-grade adenoma (Hg.ad).

^{^}14/41 families have been previously described by Ligtenberg et al, Nature Genetics, 2008 (study 4), Kovacs et al, Hum Mutat, 2009, Niessen et al, Genes, Chromosomes and Cancer, 2009, van der Klift et al, Genes, Chromosomes and Cancer, 2005 and Chan et al, Nature Genetics, 2006 (study 2).

In addition, 20 patients were reported to have constitutional mosaic methylation of the MLH1 promoter, defined as dense methylation in a proportion of MLH1 alleles or when a single nucleotide polymorphism (SNP) with three alleles (of which one is methylated) was detected. Somatic MLH1 promoter methylation is found in 85% of MMR-deficient tumors.^71–73 These sporadic MMR-deficient tumors occur in patients with a later age of onset and in the absence of a family history of CRC.^{71, 74, 75} The phenotype of patients with constitutional mosaic MLH1 promoter methylation resembles that of classical LS rather than that of somatic MLH1 promoter methylation. Of the 20 reported patients, 19 had CRC, with an average age of onset of 40.7 years for the first CRC (Table 3, Supplemental Table 3). Due to the dearth of reported cases, no significant associations can be made between the degree of mosaicism and the age of CRC onset.

The inheritance of primary epimutations is very rare due to the erasure of methylation in gametogenesis, and when it occurs, it has a non-Mendelian inheritance pattern.⁷⁶. However, germline MLH1 promoter methylation has been described in patients with early-onset CRC similar to that observed in MLH1 mutation carriers (age of onset ranging from 18–66).^77–79 Inheritance of a constitutional mosaic epimutation is expected to be even rarer, although Sloane et al. reported a family in which the parent had low-level mosaicism of MLH1 methylation and their child had non-mosaic constitutional MLH1 methylation.⁸⁰

Concluding remarks

In conclusion, mosaicism remains an underestimated cause of familial CRC and polyposis. Mosaicism present in a CRC susceptibility gene is likely to be underdiagnosed due to multiple reasons, including milder clinical phenotypes interpreted as “normal variation”, limited sensitivity of conventional molecular diagnostic techniques, limited material used for testing, and the assumption that when family history is absent, a patient’s cancer is sporadic.^{81, 82} Recent reports using methods that are able to detect germline variants with low allele frequencies, as well as variants only present in tumor material, indicate that many mosaic patients are undiagnosed. Advances in methods to evaluate the molecular biology of tumors will lead to the detection of more undiagnosed patients. Testing for mosaicism in routine diagnostics will improve counseling of these patients and allow clinical management and early detection of germline variants in the patient’s offspring even before they show a phenotype. Importantly, the studies discussed in this review demonstrate that testing tumoral DNA, rather than leukocyte DNA, will provide greater knowledge about the true incidence of mosaicism, not only in patients with APC but also in other known conditions caused by changes in CRC susceptibility genes.

Need to Know.

Background:

Somatic mosaicism is caused by genetic alterations that occur after the first division of the zygote, so that not in all cells in the body contain the same genetic variants. These variants might contribute to colorectal cancer (CRC) and polyposis syndromes.

Findings

In a systematic review, the authors found that many patients with polyposis syndromes have genetic mosaicism, most frequently in APC variants. Mosaicism in genes associated with susceptibility to CRC contributes to development of other familial CRC syndromes.

Implications for patient care:

Information on somatic mosaicism can be used in management of patients with CRC.

Abbreviations:

AFAP

Attenuated Familial Adenomatous Polyposis

CRC

Colorectal Cancer

CS

Cowden Syndrome

DHPLC

Denaturing High-Performance Liquid Chromatography

FAP

Familial Adenomatous Polyposis

HRM

High-Resolution Melting Analysis

LS

Lynch Syndrome

MMR

Mismatch Repair

PJS

Peutz-Jeghers Syndrome

PRISMA

Preferred Reporting Items for Systematic Reviews and Meta-Analyses

PPT

Protein Truncation Test