Human Genetics

, Volume 124, Issue 2, pp 105–122

Constitutional mismatch repair-deficiency syndrome: have we so far seen only the tip of an iceberg?

Authors

    • Department of Medical GeneticsMedical University of Vienna
  • Julia Etzler
    • Department of Medical GeneticsMedical University of Vienna
Review Article

DOI: 10.1007/s00439-008-0542-4

Cite this article as:
Wimmer, K. & Etzler, J. Hum Genet (2008) 124: 105. doi:10.1007/s00439-008-0542-4

Abstract

Heterozygous mutations in one of the mismatch repair (MMR) genes MLH1, MSH2, MSH6 and PMS2 cause the dominant adult cancer syndrome termed Lynch syndrome or hereditary non-polyposis colorectal cancer. During the past 10 years, some 35 reports have delineated the phenotype of patients with biallelic inheritance of mutations in one of these MMR genes. The patients suffer from a condition that is characterised by the development of childhood cancers, mainly haematological malignancies and/or brain tumours, as well as early-onset colorectal cancers. Almost all patients also show signs reminiscent of neurofibromatosis type 1, mainly café au lait spots. Alluding to the underlying mechanism, this condition may be termed as “constitutional mismatch repair-deficiency (CMMR-D) syndrome”. To give an overview of the current knowledge and its implications of this recessively inherited cancer syndrome we summarise here the genetic, clinical and pathological findings of the so far 78 reported patients of 46 families suffering from this syndrome.

Introduction

Mismatch repair (MMR) is an essential process to maintain genome integrity. It primarily corrects single base-pair mismatches and small misalignments, i.e. insertion–deletion loops (IDLs), which arise during replication. MMR increases the replication fidelity by up to three orders of magnitude. To achieve this goal (1) base–base mismatches and small IDLs have to be efficiently recognised and (2) the repair machinery has to be directed to the newly synthesised DNA strand, which carries the erroneous genetic information. The genes fulfilling these tasks were first identified in Escherichia coli where the MutS homodimer recognises mismatches and IDLs and then recruits a homodimer of MutL. This complex starts the repair process by activating the endonuclease MutH. In humans mismatch recognition is mediated by one of two heterodimers, composed of the MutS homologs MSH2 and MSH6 (MutSα) or MSH2 and MSH3 (MutSβ). MutSα which is more abundant is involved in the repair of base/base mismatches and single nucleotide misalignments while MutSβ recognises larger IDLs. Mismatch bound MutSα (or MutSβ) recruits a second heterodimer referred to as MutLα which is composed of the MutL homologs MLH1 and PMS2. In humans MutLα possesses endonuclease activity localised to the PMS2 subunit enabling MutLα to introduce random nicks at sites spanning the mismatch (Kadyrov et al. 2006). Subsequent loading of EXO1 at the 5′ side of the mismatch leads to activation of its 5′–3′ exonuclease activity resulting in removal of the incorrect DNA-fragment. The remaining single-stranded gap is filled by polymerase δ and its cofactors proliferation cell nuclear antigene (PCNA) and replication factor C (RFC). The repair process is completed when the remaining nick is sealed by DNA ligase I (reviewed by Jiricny 2006a, b).

In addition to DNA repair activity the MMR system is also involved in apoptotic response to a variety of DNA damaging agents. CMMR-deficient cells are up to 100-fold more resistant than MMR-proficient cells to apoptosis induced by O6-methylating agents and approximately 10-fold and 2-fold more tolerant to DNA damage by 6-thioguanine and cis-diammine-dichloro-platinum (also known as cisplatin), respectively (reviewed by Stojic et al. 2004). Although the precise mechanisms involved remain to be elucidated it is believed that by recognition of DNA damage and repeated unsuccessful attempts of DNA repair the MMR system causes replication blocks which in turn induces cell-cycle arrest. However, there is also evidence that MMR proteins play a role in direct apoptotic signalling (reviewed by Jiricny 2006a).

The contribution of defective mismatch repair (MMR) to the development of human cancer has been recognised for more than a decade (for review see Peltomaki 2003). Most importantly, heterozygous inactivating germ line mutations in the MMR genes MLH1, MSH2, MSH6 and PMS2 cause a cancer predisposition syndrome named Lynch syndrome (LS) or hereditary non-polyposis colorectal cancer (HNPCC) which is characterised by early onset of colorectal cancer (CRC) and a highly increased risk for endometrial cancer in women (Lynch and de la Chapelle 2003; Watson and Lynch 1994). Cancers of the stomach, the small bowel, the ureter and the ovary are also associated with this disorder. Tumours arising in LS patients result from somatic loss of the remaining wild type MLH1, MSH2, MSH6 or PMS2 allele which leads to impaired MMR and accumulation of somatic mutations in cancer genes as well as in repetitive sequence motives termed microsatellites. Shortening and lengthening of microsatellites due to unrepaired IDLs in these elements is a hallmark of tumours that arise due to impaired MMR and is referred to as microsatellite instability (MSI). Up to 90% of the pathogenic alleles causing LS have been found in MSH2 and MLH1, roughly 10% in MSH6 and only a small minority in PMS2 (Peltomaki 2001). Furthermore, epigenetic silencing of MLH1 by promoter hypermethylation causes defective MMR in approximately 10–15% of sporadic (non-familial) colorectal cancers (CRC) (Herman et al. 1998).

While individuals with the autosomal dominant LS harbour heterozygous mutant MMR gene alleles, there are now up to 35 reports of patients that suffer from a novel autosomal recessively inherited cancer syndrome that is caused by biallelic germ line mutations in one of the MMR genes. The patients develop childhood malignancies, mainly haematological malignancies and/or brain tumours as well as very early-onset CRC. Almost all patients also show some signs reminiscent of neurofibromatosis type 1 (NF1), mainly café au lait spots (CLS).

In this article we will summarise the genetic, clinical and pathological findings of 78 patients from 46 families that have been reported to suffer from this syndrome which was firstly described in 1999 in two simultaneously published reports (Ricciardone et al. 1999; Wang et al. 1999) and since then has received several names including childhood cancer syndrome (CCS) (Kruger et al. 2008), Lynch III syndrome (Felton et al. 2007b), or the acronym “CoLoN”, Colon tumours or/and Leukaemia/Lymphoma or/and Neurofibromatosis features (Bandipalliam 2005). We will here refer to it as constitutional MMR-deficiency (CMMR-D) syndrome. This term is an extension of the name MMR-D syndrome given by Scott et al. (2007b). It alludes to the underlying mechanism of this cancer syndrome and clearly refers to a constitutional condition avoiding possible confusion with somatic MMR-D as seen in tumours of LS patients due to somatic inactivating mutations and in tumours of sporadic CRC patients with MLH1 hypermethylation. Therefore, CMMR-D syndrome is in our opinion currently the most precise and at the same time most comprehensive name for the syndrome.

The tumour spectrum in CMMR-D syndrome

The clinical presentation and genetic findings of 78 individuals with homozygous or compound heterozygous mutations in one of the MMR genes are summarised in Table 1. Fourteen patients of nine families had biallelic mutations in the MLH1 gene. Biallelic MSH2 and MSH6 mutations were found in eight patients of four families and in thirteen patients of nine families, respectively. The largest group consisted of 43 patients from 24 families carrying biallelic PMS2 mutations. This group included ten patients of Pakistani origin that were homozygous for a founder PMS2 mutation (De Vos et al. 2006). In total, these patients developed 111 malignancies that are listed in Table 2. The spectrum of tumours can largely be divided in four groups: (1) haematological malignancies, (2) brain tumours, (3) LS-associated tumours and (4) others. The overall picture arising from this list of tumours shows that haematological malignancies usually arise in infancy or early childhood (mean age of onset 5.5 years), while brain tumours develop somewhat later in childhood of CMMR-D syndrome patients (mean age of onset 8 years). Tumours belonging to the LS spectrum, primarily CRC, usually arise in adolescent or young adult CMMR-D syndrome patients (mean age of onset 16 years) and can often be found as second or even third malignancy in patients who survived their first tumour [19/41 (46%) of CMMR-D syndrome patients with tumours belonging to the LS spectrum].
Table 1

List of patients with CMMR-D syndrome

Family

Gene

Mutations

Family history

Patients

Malignant tumours (age in years at diagnosis or at death of disease)

Signs of neurofibromatosis

Results of microsatellite instability (MSI) analysis

Results of immunohistochemical analysis

Reference

1

MLH1

c.[131C>T]+[1321G>A], p.[Ser44Phe]+[Ala441Thr]

Mother (p.Ser44Phe) of LS-fam., father (p.Ala441Thr) breast ca. (65), 2°pat CRC

Pat.1

Breast ca. (35)

NR

MSI present in tumour and lymphocytes*

NA

Hackman et al. 1997

2

MLH1

c.[676C>T]+[676C>T], p.[Arg226X]+[Arg226X]

Consang. LS-fam.

Pat.1

Acute leukaemia (2)

NR

NA

NA

Ricciardone et al. 1999

Pat.2

NHL (3.25)

CLS

NA

NA

Pat.3

Atypical CML (1)

CLS, 2x fibromatous skin tumours

NA

NA

3

MLH1

c.[199G>T]+[199G>T], p.[Gly67Trp]+[Gly67Trp]

Consang. LS-fam.

Pat.1

NHL (2)

CLS, tibia-pseudarthrosis

NA

NA

Wang et al. 1999

Pat.2

AML (6), medulloblastoma (7)

CLS, multiple nfbs.

MSI present in normal buccal mucosa cells*

NA

4

MLH1

c.[1852_1853AA>GC]+[546-2A>G], p.[Lys618Ala]+aberrant splicing

c.546-2A>G segregating with LS in the family

Pat.1

CRC (35), sarcoma (65)

NR

MSI present in CRC and sarcoma, MSI absent in lymphocytes*

NA

Liu et al. 1999

5

MLH1

c.[1732-?_1896+?del]+[1732-?_ 1896+?del], p.[Glu578_Glu632del]+p.[Glu578_ Glu632del]

Both parents of LS-fam.

Pat.1

Glioma (4)

CLS, axillary freckling

MSI present in normal autopsy tissue*

NA

Vilkki et al. 2001

6

MLH1

c.[104_105TG>AC]+[596_597delAG] p.[Met35Asn]+[Glu199AspfsX4]

Mother (p.Glu199AspfsX4) of LS-fam., 2° pat (p.Met35Asn) CRC (42)

Pat.1

Glioblastoma (4), Wilms tumour (4)

CLS

MSI present in Wilms tumour, MSI absent in glioblastoma

MLH1/MSH2/MSH6/PMS2: −/+/+/− in glioblastoma, Wilms tumour and non-neoplastic cells

Wagner et al. 2003; Poley et al. 2007

7

MLH1

c.[1942C>T]+[1942C>T], p.[Pro648Ser]+[Pro648Ser]

Consang. LS-fam.

Pat.1

No malignant tumours (6)

CLS, nfb.

NA

NA

Raevaara et al. 2004

8

MLH1

c.[2059C>T]+[2059C>T], p.[Arg687Trp]+[Arg687Trp]

Consang. fam., 2° CRC (60–70), 2° gastric ca. (58)

Pat.1

Duodenal ca. (11)

CLS

MSI present in duodenal ca. liver metastasis and lymphoblasts*

MLH1/MSH2/MSH6: +/+/+

Gallinger et al. 2004

Pat.2

3x malignant colonic adenomatous polyps (9)

CLS, axillary freckling, Lisch nodule

MSI present in malignant colon polyps and lymphoblasts*

MLH1/MSH2/MSH6: +/+/+

Pat.3

No malignant tumours (7)

CLS, plexiform nfb. of the tongue, Lisch nodule, (hairy nevi)

MSI absent in plexiform nfb., MSI present in lymphoblasts*

MLH1/MSH2/MSH6: +/+/+

9

MLH1

c.[806C>G]+[806C>G], p.[Ser269X]+[Ser269X]

Both parents of LS-fam., founder effect possible

Pat.1

CRC (22)

NR

NA

NA

Rey et al. 2004

10

MSH2

c.[1662-1G>A]+[1662-1G>A], p.[Ser554_Gly587>ArgfsX11]+[Ser554_Gly587>ArgfsX11]

No family history of LS-related cancers, maybe due to young age of the parents and grandparents

Pat.1

T-cell ALL (2), B-cell lymphoma (8)

CLS, (IgA deficiency)

MSI present in normal blood lymphocytes*

NA

Whiteside et al. 2002; Felton et al. 2007a

11

MSH2

c.[1-?_1076+?del]+[454delA], p.[Met1_Arg359>IlefsX29]+[Met152CysfsX22]

3°mat (p.Met152CysfsX22) EC (43), 3°pat

Pat.1

T-cell lymphoma (1.25)

NR

NA

NA

Bougeard et al. 2003

(p.Met1_Arg359>IlefsX29) astrocytoma (27), 3°pat EC (59)

Pat.2

Glioblastoma (3)

NR

MSI absent in glioblastoma

NA

12

MSH2

c.[2006-5T>A; 2006-6T>A]+[2006-5T>A; 2006-6T>A], p.[Pro670_Arg737>LeufsX7 (+) Thr668_Gly669insVal] (two different splicing effects)

Consang. fam., no LS-related cancers reported

Pat.1

2x CRC, (multiple polyps in duodenum, colon and rectum) (11)

CLS

MSI present in CRC

MLH1/MSH2/MSH6/PMS2: +/−/−/+

Muller et al. 2006

Pat.2

3×CRC, (multiple polyps in duodenum and colon) (12)

CLS

MSI present in CRC

MLH1/MSH2/MSH6/PMS2: +/−/−/+

13

MSH2

c.[226C>T]+[226C>T], p.[Gln76X]+[Gln76X]

Consang. LS-fam.

Pat.1

T-cell NHL (0.4)

CLS, (hypopigmented skin areas)

NA

NA

Scott et al. 2007a

Pat.2

T-cell NHL (2.5), (multiple colonic adenomas) (6)

CLS, (hypopigmented skin areas)

NA

NA

Pat.3

T-cell NHL (2.5)

CLS, (hypopigmented skin areas)

NA

NA

14

MSH6

c.[3386_3388delGTG]+[3386_3388delGTG] and p.[Cys1129_Val1130>Leu]+[Cys1129_Val1130>Leu]

Consang., 2°mat. CRC (47)

Pat.1

Oligodendroglioma (10), CRC (12)

CLS

MSI absent in oligodendroglioma, MSI present in CRC

MLH1/MSH2/MSH6: +/+/+ in oligodendroglioma, MLH1/MSH2/MSH6: +/+/− in CRC

Menko et al. 2004

15

MSH6

c.[3635dupT]+[3635dupT], p.[Asp1213GlyfsX2]+[Asp1213Gly fsX2]

3°mat CRC (48)

Pat.1

Lymphoma (5), CRC (8)

CLS, axillary freckling

MSI present in glioblastoma, MSI absent in normal blood lymphocytes

 

Hegde et al. 2005

Pat.2

Glioblastoma multiforme (8)

CLS, axillary freckling

16

MSH6

c.[3020G>A]+[3607_3610delCATG], p.[Trp1007X]+[His1203_Ala1204>GlnfsX12]

Mother (p.Trp1007X) of LS-fam.

Pat.1

Astrocytoma (9), T-cell lymphoma (10)

CLS, axillary freckling, (IgA deficiency)

NA

MLH1/MSH2/MSH6: +/+/− in astrocytoma

Ostergaard et al. 2005

Pat.2

Glioblastoma on spinal cord (2)

CLS, axillary freckling, (IgA deficiency)

NA

MLH1/MSH2/MSH6: +/+/− in glioblastoma

17

MSH6

c.[3226C>T]+[3991C>T], p.[Arg1076Cys]+[Arg1331X, Ala1268_Arg1334>GlyfsX6]

2°pat (c.3991C>T) CRC (73)

Pat.1

CRC (19), EC (24)

CLS, (lupus erythematosus)

MSI present in CRC

MLH1/MSH2/MSH6: +/+/− in CRC, normal colonic mucosa and lymphocytes

Plaschke et al. 2006

18

MSH6

c.[2295C>G]+[2633T>C], p.[Cys765Trp]+[Val878Ala]

Mother (p.Cys765Trp) EC (60)

Pat.1

CRC (31)

 

NA

NA

Plaschke et al. 2006

19

MSH6

c.[642C>G]+[458-1G>A], p.[Tyr214X]+[?] (aberrant splicing)

NR

Pat.1

Medulloblastoma (7), AML (10), 2x CRC, (multiple colonic adenomas) (13)

CLS, (hairy nevi, hypopigmented skin areas, IgA and IgG2 deficiency)

NA

MLH1/MSH2/MSH6: +/+/− in colonic adenoma

Scott et al. 2007b

20

MSH6

c.[1596dupT]+[3261delC], p.[Ser532PhefsX1]+[Phe1088SerfsX2]

3°mat (p.Phe1088SerfsX2) EC (59)

Pat.1

Multiple colonic polyps with high degree of dysplasia (9)

CLS, Lisch nodules

Low MSI present in adenoma, MSI absent in lymphocytes*

MSH6: − in neoplastic and normal colon epithelial tissue

Auclair et al. 2007

Pat. 2

Glioblastoma (7)

CLS

NA

NA

21

MSH6

c.[4002-31_4002-8delins24]+[4002- 31_4002-8delins24], 2 aberrantly spliced transcripts

Consang., 2°mat. gastric ca.

Pat.1

Medulloblastoma (6), MDS/AML (9)

CLS

MSI absent glioblastoma and normal blood lymphocytes

MLH1/MSH2/MSH6/PMS2: +/−/−/+ in glioblastoma, +/+/−/+ in non-neoplastic cells

Etzler et al. 2008

Pat. 2

Glioblastoma multiforme (9)

CLS, (hypopigmented skin areas)

22

MSH6

c.[1806_1809delAAAG]+[3226C>T], p.[Glu604LeufsX5]+[Arg1076Cys]

No family history of LS-related cancers

Pat.1

4x CRC, (3x colonic adenomas) 17

CLS absent, (vitiligo, lupus erythematosus)

MSI present in CRC

MLH1/MSH2/MSH6/PMS2: +/+/−/+ in CRC and normal colonic mucosa

Rahner et al. 2008

23

PMS2

c.[1221delG (+) 2361_2364 delCTTC], p.[Thr408LeufsX40 (+) Phe788_Met789>CysfsX2]

2°mat. (c.1221delG) unspecified cancer (63)

Pat.1

Oligodendroglioma (14), CRC, (recto-sigmoid polyp) (18)

NR

MSI present in CRC and normal colonic mucosa*

NA

De Rosa et al. 2000

Pat.2

Neuroblastoma (13)

NR

24

PMS2

c.[1169_1170ins20]+[1169_1170ins20], truncating mutation

Consang., 2°pat CRC (53)

Pat.1

2x CRC, (multiple colonic adenomas) (16), ovarian neuroectodermal tumour (21), EC (23), brain tumour (24)

CLS

MSI present in CRC and ovarian neuroectodermal tumour

NA

Trimbath et al. 2001

Pat.2

Anaplastic astrocytoma (7), (3x colonic adenomatous polyps (20))

CLS

NA

NA

Pat.3

ALL (4)

CLS

NA

NA

25

PMS2

c.[2404C>T]+[2404C>T], p.[Arg802X]+[Arg802X]

Consang., no family history of LS-related cancers

Pat.1

B-cell NHL (10)

CLS

NA

NA

De Vos et al. 2004

Pat.2

SPNET (8)

CLS

NA

NA

Pat.3

SPNET (14)

CLS

NA

NA

Pat.1B

No malignant tumours (10)

CLS

NA

NA

26

PMS2

c.[400C>T]+[2184_2185delTC], p.[Arg134X]+[Leu729GlnfsX6]

NR

Pat.1

Glioblastoma (4), (2x colonic adenomas (13)), NHL of the rectum (17), glioblastoma (21)

CLS

MSI present in both glioblastomas and colonic adenoma

NA

Hamilton et al. 1995; Taylor et al. 1999; De Vos et al. 2004

Pat.2

CRC, (3x colonic adenomas) (11), (multiple colonic polyps (14))

CLS

MSI present in CRC

NA

27

PMS2

c.[137G>T (+) 1927C>T], p.[Ser46Ile (+) Gln643X]

Mother (c.1927C>T) dysplastic rectal adenoma (44), 2°mat jejunal ca. (52), father lung ca. (55), 2°pat CRC

Pat.1

Duodenal ca., (4x colonic adenomas) (16), glioblastoma (17)

CLS

MSI present in duodenal ca., MSI absent in glioblastoma, MSI present in normal tissue*

MLH1/MSH2/MSH6/PMS2: +/+/+/− in all tumour tissues, both in neoplastic and in normal cells

Agostini et al. 2005

28

PMS2

c.[2404C>T]+[2404C>T], p.[Arg802X]+[Arg802X]

Consang., no family history of LS-related cancers

Pat.1

T-cell leukaemia (2), T-cell lymphoma (14), multiple CRCs (18)

CLS

NA

NA

De Vos et al. 2006

29

PMS2

c.[543delT]+[543delT], p.[Tyr181X]+[Tyr181X]

Consang., no family history of LS-related cancers

Pat.1

SPNET (8)

CLS

NA

NA

De Vos et al. 2006

Pat.2

SPNET (4)

CLS

NA

NA

 

30

PMS2

c.[2404C>T]+[2404C>T], p.[Arg802X]+[Arg802X]

Consang., no family history of LS-related cancers

Pat.1

Glioma (15)

CLS

NA

NA

De Vos et al. 2006

Pat.2

ALL (15)

CLS

NA

NA

Pat.3

Astrocytoma (6), glioblastoma (7)

CLS

NA

NA

31

PMS2

c.[2404C>T]+[2404C>T], p.[Arg802X]+[Arg802X]

Consang., no family history of LS-related cancers

Pat.1

Glioblastoma (2)

NR

NA

NA

De Vos et al. 2006

Pat.2

T-cell NHL (3)

NR

NA

NA

Pat.3

ALL (6)

NR

NA

NA

32

PMS2

c.[2404C>T]+[2404C>T], p.[Arg802X]+[Arg802X]

Consang., no family history of LS-related cancers

Pat.1

ALL (6)

CLS

NA

NA

De Vos et al. 2006

33

PMS2

c.[137G>T]+[1730dupA;1732C>T], p.[Ser46Ile]+[Arg578ValfsX3]

2°mat (c.1730dupA; 1732C>T) CRC (73), 2x 3°mat CRC (>65)

Pat.1

Oligodendroglioma (19), 2x CRC, (colonic adenomatous polyp) (24)

NR

 

Western blot: PMS2 absent in lymphoblastoid cell line

Auclair et al. 2007

Pat.2

CRC, (multiple adenomatous polyps) (20), EC (24)

CLS, brain angioma (2)

MSI present in CRC

MLH1/MSH2/MSH6/PMS2: +/+/+/− in EC, +/+/−/− in CRC, Western blot: PMS2 absent in lymphoblastoid cell line

34

PMS2

c.[1768delA]+[1768delA], p.[Ile590PhefsX5]+[Ile590PhefsX5]

Consang., no family history of LS-related cancers

Pat.1

Glioblastoma multiforme (6), CRC, jejunum ca., ureter/renal pelvis ca., (multiple colonic polyps) (15)

CLS

MSI present in CRC

MLH1/MSH2/MSH6/PMS2: +/+/+/− in CRC and in adjacent non-neoplastic cells

Kruger et al. 2008

Pat.2

Glioblastoma multiforme (6)

CLS

Pat.3

(3x colonic polyps (8)), glioblastoma multiforme (9)

NR

Pat.4

Infantile myofibromatosis of the neck (1)

NR

35

PMS2

c.[812G>T]+[812G>T], p.[Gly271Val]+[Gly271Val]

Consang., 3°mat EC (61), 3°mat abdominal ca. (75), two more cancers of unknown site and unknown age

Pat.1

2x CRC, (multiple colonic polyps) (13, 14)

CLS, freckling

MSI present in CRC

MLH1/MSH2/MSH6/PMS2: +/+/+/− in CRC and in adjacent non-neoplastic cells

Kruger et al. 2008

Pat 2

T-cell NHL (10), CRC, (multiple s colonic polyps) (11)

CLS

36

PMS2

c.[906-?_2589+?del]+[906-? _2589+?del], p.[Val302_X863 del]+[Val302_X863del] deletion involves PMS2 (exons 9-15), oncomudulin, TRIAD3 and FSCN1

Consang., no family history of LS-related cancers

Pat.1

10x CRC, (multiple adenomatous polyps) (23), duodenal ca. (25)

CLS, dysmorphic features, mental retardation

MSI present in most CRCs and in adenomas, MSI absent in normal tissue

PMS2: − in all tumours and normal tissue

Will et al. 2007

37

PMS2

c.[1306dupA]+[1306dupA], p.[Ser436LysfsX22]+[Ser436LysfsX22]

Consang., no family history of LS-related cancers

Pat.1

T-cell NHL (6), multiple CRCs, (multiple colonic adenomas) (16)

CLS

MSI present in CRC

MLH1/MSH2/MSH6/PMS2: +/+/+/−

Kratz et al. 2007

Pat 2

SPNET (9)

CLS

38

PMS2

c.[182delA]+[182delA], p.[Tyr61LeufsX15]+[Tyr61LeufsX15]

Consang., no family history of LS-related cancers

Pat.1

Glioblastoma (10)

CLS

MSI absent in normal blood lymphocytes

MLH1/MSH2/MSH6/PMS2: +/+/+/− in glioblastoma and normal tissue

Etzler et al. 2008

39

PMS2

c.[706-?_803+?del]+[706-?_803 +?del], p.[Leu236_Tyr268>HisfsX30]+[Leu236_Tyr268>HisfsX30]

Consang., mother rectal adenoma, multiple cases of cancer in the family: PNET (10), T-cell lymphoblastic leukaemia (2), salivary gland ca. (20) CRC (66)

Pat.1

Multiple CRCs, (multiple colonic adenomas) (15)

CLS

 

MLH1/MSH2/MSH6/PMS2: +/+/+/− in CRC and normal tissue

Tan et al. 2008

Pat.2

Glioblastoma (8)

CLS

40

PMS2

c. [137G>A]+[137G>T], p.[Ser46Asn]+[Ser46Ile]

Multiple cases of LS-related cancers in mat (p.Ser46Ile) family

Pat.1

CRC, (multiple adenomatous polyps) (14)

CLS

MSI present in CRC

MLH1/MSH2/MSH6/PMS2: +/+/+/− in CRC and normal colonic mucosa

Jackson et al. 2008

41

PMS2

c.[219T>A]+[219T>A], p.[Cys73X]+[Cys73X]

Consang., no family history of LS-related cancers

Pat.1

Rhabdomyosarcoma (4), CRC (9)

CLS

  

Kratz and Wimmer, unpublished

42

PMS2

c.[2249G>A]+[1-?_2589+?del], p.[Gly750Asp]+[Met1_X863del] (complete gene loss)

Bother CRC (21)

Pat.1

CRC (22), brain tumour (23)

NR

 

PMS2 absent in CRC and adjacent healthy tissue

Senter et al. 2008

43

PMS2

c.[904-?_1144+?del]+[1A>G], p.[Cys303_Asn383>ThrfsX2]+[?] (5′ truncation)

2 x 4° CRC (48, 56)

Pat.1

Multiple CRCs (28)

NR

 

PMS2 absent in CRC and adjacent healthy tissue

Senter et al. 2008

44

PMS2

c.[614A>C]+[1A>G], p.[Gln205Pro]+[?] (5′ truncation)

Brother brain tumour (38), sister brain tumour (31)

Pat.1

CRC (20), duodenal ca. (41), lymphoma (?)

NR

 

PMS2 absent in CRC and adjacent healthy tissue

Senter et al. 2008

45

PMS2

c.[251-2A>G]+[1A>G], p.[Leu85_118Ser>MetfsX17] (loss of exon 3 splice acceptor site)+[?] (5′ truncation)

Brother CRC (26), glioblastoma (34), brother glioma (24)

Pat.1

CRC (24), EC (35), glioma (35)

NR

 

PMS2 absent in CRC and adjacent healthy tissue

Senter et al. 2008

46

PMS2

c.[949C>T]+[949C>T], p.[Gln317X]+[Gln317X]

Consang., brother T-cell lymphoma (6)

Pat.1

Medulloblastoma (7), CRC (16)

Present (not specified)

 

PMS2 absent in CRC and adjacent healthy tissue

Senter et al. 2008

Family history: LS Lynch syndrome, fam. family, ca. cancer, CRC colorectal cancer, pat paternal, consang. consanguineous, mat maternal, EC endometrial cancer, PNET primitive neuroectodermal tumour

Patients: pat. patient

Malignant tumours: ca. cancer, NHL non-Hodgkin lymphoma, CML chronic myelogenous leukaemia, AML acute myelogenous leukaemia, CRC colorectal cancer, ALL acute lymphoblastic leukaemia, EC endometrial cancer, MDS myelodysplastic syndrome, SPNET supratentorial primitive neuroectodermal tumour

Signs of neurofibromatosis: NR not reported, CLS café au lait spots, nfb. neurofibroma

Results of microsatellite instability analysis: MSI microsatellite instability, NA not analysed, CRC colorectal cancer

Results of immunohistochemical analysis: NA not analysed, + positive staining, negative staining, CRC colorectal cancer

* small pool PCR was applied

Table 2

List of malignancies in CMMR-D syndrome patients

Malignancies

No. of tumours

Median age at diagnosis (range) (years)

Haematological malignancies

 ALL

7 (6%)

4 (2–15)

 AML

3 (3%)

9 (6–10)

 Lymphomas

16 (14%)

5 (0.4–17)

 Atypical CML

1 (1%)

1

 Total

27 (24%)

5.5 (0.4–17)

Brain tumours

 Glioblastoma and other astrocytic tumours

25 (23%)

8 (2–35)

 SPNET

5 (5%)

8 (4–14)

 Medulloblastoma

4 (4%)

7 (6–7)

 Unspecified

2 (2%)

23 and 24

 Total

36 (32%)

8 (2–35)

LS-associated tumours

 CRC

31 (28%)

16 (8–35)

 Endometrial ca.

4 (4%)

24 (23–35)

 Duodenum/jejunum ca.

5 (5%)

16 (11–41)

 Ureter/renal pelvis ca.

1 (1%)

15

 Total

41 (37%)

16 (8–41)

Others

 Neuroblastoma

1 (1%)

13

 Wilms tumour

1 (1%)

4

 Ovarian neuroectodermal tumour

1 (1%)

21

 Infantile myofibromatosis

1 (1%)

1

 Rhabdomyosarcoma

1 (1%)

4

 Mamma ca.

1 (1%)

35

 Sarcoma

1 (1%)

65

ALL acute lymphoblastic leukaemia, AML acute myelogenous leukaemia, CML chronic myelogenous leukaemia, SPNET supratentorial primitive neuroectodermal tumour, CRC colorectal cancer, ca. carcinoma

The tumour spectrum and the age of onset in MLH1/MSH2 versus MSH6 and PMS2 mutation carriers, respectively, are compared in Table 3. In general, CMMR-D syndrome patients carrying MLH1/MSH2 mutations have an earlier age of malignancy onset (mean age 3.5 years) than patients with biallelic MSH6 and PMS2 mutations, respectively, (mean age 9 years). The prevalence of haematological tumours is higher in patients with MLH1 or MSH2 mutations, whereas patients with biallelic MSH6 and PMS2 mutations have a higher incidence of brain and LS-associated tumours (Table 3). The probability to survive the first tumour and to develop a different second or third malignancy is higher in patients with MSH6 or PMS2 (23/55; 42%) than in patients with biallelic MLH1 or MSH2 mutations (3/20; 15%). These differences are even more pronounced when families lacking patients with haematological malignancies, brain tumours and/or embryonal cancers are excluded from the comparison. This group comprises eleven patients of nine families, i.e. families 8, 9, 12, 17, 18, 22, 36, 40 and 43 in Table 1, who developed only LS-associated tumours at a mean age of 17 years, one patient who developed breast cancer at the age of 35 years (Hackman et al. 1997) and one patient with a CRC and a sarcoma at 35 and 65 years, respectively (Liu et al. 1999). Twelve of the 22 pathogenic mutations (54%) in these eleven families were missense mutations that possibly code for proteins with some residual functionality. This number is substantially larger than the overall frequency of missense mutations, i.e. 27% (25/92), in the so far reported CMMR-D syndrome families. Hence, the “milder” phenotype in the families who lack patients with haematological malignancies and/or brain tumours in early childhood might—at least partly—be explained by some residual MMR proficiency.
Table 3

Differences in tumour spectrum and age of malignancy onset between MLH1/MSH2 and MSH6/PMS2 mutation carriers

Gene

Patients (families)

Haematological malignancies

Brain tumours

LS-associated malignancies

Others

Median age at diagnosis of primary tumour (years)

MLH1

12 (8)

5 (42%)

3 (25%)

4 (33%)

3 (25%)

5 (1–35)

MSH2

8 (4)

6 (75%)

1 (12.5%)

2 (25%)

2.5 (0.4–12)

MLH1/MSH2

20 (12)

11 (55%)

4 (20%)

6 (30%)

3 (15%)

3.5 (0.4–35)

MSH6

13 (9)

4 (31%)

8 (61.5%)

8 (61.5%)

9 (2–31)

PMS2

42 (24)

12 (29%)

24 (57%)

27 (64%)

4 (9.5%)

9.5 (1–28)

MSH6/PMS2

55 (33)

16 (29%)

32 (58%)

35 (64%)

4 (7%)

9 (1–31)

In general, heterozygous MLH1 and MSH2 mutations are associated with a higher penetrance of LS than heterozygous MSH6 mutations (Plaschke et al. 2004) and, especially, PMS2 mutations (Senter et al. 2008; Truninger et al. 2005). Thus, it is not surprising that none of the families with CMMR-D syndrome patients due to biallelic PMS2 mutations had a clear family history of LS, although some of them showed an increased tumour incidence. The trend towards a more severe phenotypic expression of MLH1/MSH2 than MSH6 and, particularly, PMS2 mutations appears to hold true also for CMMR-D syndrome. While most of the children with biallelic MLH1 and MSH2 mutations die from haematological malignancies or glioblastoma early in childhood, there is a tendency towards later onset brain and LS-associated tumours in patients with biallelic MSH6 and PMS2 mutations. Exceptions to this general rule may be patients carrying MLH1 and MSH2 missense alleles with some residual functionality.

Brain tumours and haematological malignancies in CMMR-D syndrome

The most prevalent brain tumours in CMMR-D syndrome patients are astrocytomas, primarily glioblastomas. Five patients developed supratentorial primitive neuroectodermal tumours (SPNET) and all of them carried homozygous PMS2 mutations. Four of these patients were from consanguineous Pakistani families (De Vos et al. 2004, 2006) and the fifth patient was the deceased brother of an index patient from consanguineous Turkish parents in whom diagnosis was inferred (Kratz et al. 2007). Little is known on the molecular mechanisms underlying this rare brain tumour entity and the association with PMS2 is intriguing. Therefore, it will be interesting to learn whether CMMR-D syndrome associated SPNET is restricted to patients with PMS2 deficiency and to determine whether defective MMR is a frequent finding in this tumour entity.

Four CMMR-D syndrome patients developed medulloblastoma (Etzler et al. 2008; Scott et al. 2007b; Senter et al. 2008; Wang et al. 1999). Three of them developed also acute myeloid leukaemia (AML) and are so far the only CMMR-D patients with this haematological malignancy. The most prevalent haematological cancers in CMMR-D patients are (non-Hodgkin) lymphoma and acute lymphoblastoid leukaemia, mainly of the T-cell type. The observation of AML in CMMR-D syndrome raises two questions.

Firstly, since AML was the second malignancy in two of the AML patients (Etzler et al. 2008; Scott et al. 2007b), it is possible that the AML in these patients was therapy related. Partial loss of chromosome arm 7q due to a translocation t(2;7)(p13;q21) in the leukaemic cells of one of them supports this possibility (Etzler et al. 2008), since monosomy 7 or deletion of 7q is a common finding in therapy-related MDS and AML (Le Beau et al. 1986; Rowley et al. 1981; Smith et al. 2003; Thirman and Larson 1996). However, the fact that in the third patient AML was the primary malignancy argues against this possibility (Wang et al. 1999). There is evidence, mainly from in vitro data, that MMR deficiency may confer resistance to certain chemotherapeutics and may increase their mutagenic potential. Therefore, Scott et al. (2007b) suggested that the avoidance of chemotherapeutics, such as O6-methylating agents, may be considered when delineating treatment approaches for patients with this special genetic background.

Secondly, myeloid leukaemia arising in patients with multiple Café au lait spots (CLS) is usually considered highly suggestive for NF1 and, therefore, may preclude proper diagnosis of CMMR-D syndrome. Children with NF1 have an increased risk to develop juvenile myelomonocytic leukaemia (JMML), a rare mixed myelodysplastic/myeloproliferative disorder. Involvement of NF1 inactivation in the pathogenesis of this disorder has clearly been established (Shannon et al. 1994). Moreover, preleukaemia with bone marrow monosomy 7 and acute myelogenous leukaemia (AML) has also been reported to be associated with NF1 by Maris et al. (1997) and Perilongo et al. (1993). In these two reports all together seven—albeit genetically unconfirmed—NF1 patients who developed different primary malignancies and secondary therapy-related MDS with monosomy 7 were described. In none of five tested patients LOH of the NF1 locus was found although this is the most frequent second hit mechanism of NF1 inactivation in myeloid malignancies (Stephens et al. 2006). Two of the patients were siblings with evidence of possible parental consanguinity (Maris et al. 1997). The primary malignancies in these seven patients included brain tumours and lymphoblastic leukaemias as well as Wilms tumour and rhabdomyosarcoma. The latter were each so far seen only once in CMMR-D patients (Poley et al. 2007; Wagner et al. 2003; Kratz and Wimmer unpublished results). As NF1 mutation analysis was not performed in these cases, it is conceivable that some of the above mentioned patients suffered from CMMR-D, rather than NF1. Taken the overlap of clinical signs of CMMR-D syndrome and NF1, we expect that testing all patients with signs reminiscent of NF1 and malignancies other than clearly NF1-associated tumours for biallelic mutations in one of the MMR genes will clarify the association of these malignancies with NF1 and/or CMMR-D syndrome.

NF1 features in CMMR-D patients

NF1 is an autosomal dominantly inherited disease, but approximately 50% of NF1 patients are sporadic patients carrying de novo mutations. The main features of NF1 are CLS, skin-fold freckling, iris Lisch nodules and neurofibromas. Less frequent complications of NF1 include specific bone lesions such as tibial pseudarthrosis and a number of benign or malign tumours including optic gliomas or pilocytic astocytomas at other locations and JMML. CLS are most often the first sign of NF1 usually arising in the first two years of life and most infants with more than six CLS (diameter >0.5 cm) will later in life develop more signs of NF1 (Korf 1992). Therefore, multiple CLS in young children are highly suggestive for NF1 although they are not pathognomonic for NF1. According to the NIH consensus conference statement (Stumpf 1988) the patient has to show at least one additional criterion before the diagnosis can be made on clinical grounds. Of the so far 78 described cases with CMMR-D 57 are reported to show multiple CLS and 17 would have fulfilled the criteria for clinical diagnosis of NF1. Therefore, it is not surprising that some of these patients were initially misdiagnosed as NF1 patients. However, in none of the so far tested CMMR-D syndrome patients a pathogenic NF1 germ line mutation could be identified (Auclair et al. 2007; Etzler et al. 2008; Hegde et al. 2005; Menko et al. 2004; Ostergaard et al. 2005; Trimbath et al. 2001). Furthermore, several reports stress that CLS in patients with CMMR-D differ from typical NF1-associated CLS in that they often vary in their degree of pigmentation and have irregular borders (De Vos et al. 2006; Kruger et al. 2008; Scott et al. 2007a, b; Tan et al. 2008). Nevertheless, it is tempting to speculate that NF1 mutations are responsible for the formation of CLS and other signs reminiscent of NF1 in CMMR-D syndrome patients. It is possible that CMMR-D syndrome patients may carry a mosaic NF1 mutation which has occurred in an early developmental stage. This notion is supported by a segmental hemicorporal distribution of CLS reported in some of these patients (Auclair et al. 2007) and cell culture studies that render evidence that NF1 is a mutational target of MMR deficiency (Wang et al. 2003). However, further direct evidence has to be awaited to confirm or refute this possibility. CLS are also found in the context of other hereditary syndromes, e.g. NF1-like syndrome (SPRED1 gene) (Brems et al. 2007), LEOPARD syndrome, Nijmegen breakage syndrome (NSB gene) and Fanconi anaemia subtype-D1 (FA-D1) which shows clinical overlap with CMMR-D syndrome and is caused by biallelic BRCA2 mutations (Hirsch et al. 2004; Reid et al. 2005). Hence, it is conceivable that also other genetic alterations may account for the development of CLS.

The overlap of CMMR-D syndrome with NF1 has also implications for clinical diagnosis of NF1. In particular, the clinical criterion of “first-degree relative” has to be reconsidered in view of the appearance of NF1 signs in equally affected siblings with CMMR-D syndrome. Therefore, it has been suggested to change the wording from “first-degree relative” (which includes siblings) to “parent or offspring” in the NIH diagnostic criteria for NF1 (Huson 2008).

Taken together, multiple CLS is a very common feature, if not a hallmark, of CMMR-D syndrome and we suggest that any child presenting with multiple CLS and an early-onset not clearly NF1-associated malignancy should be suspected and tested for CMMR-D syndrome.

Other cutaneous and immunological features in CMMR-D patients

So far there is only one compound heterozygous MSH6 mutation carrier who is explicitly reported to lack CLS or other signs of NF1 (Rahner et al. 2008). This patient as well as five additional patients (Etzler et al. 2008; Scott et al. 2007a, b) presented with hypopigmented skin areas which may be an additional or alternative cutaneous feature indicative of CMMR-D syndrome. Further, two patients compound heterozygous for the MSH6 missense mutation p.Arg1076Cys developed Lupus erythematosus (Plaschke et al. 2006; Rahner et al. 2008). IgA deficiency is a feature reported so far in one patient with a homozygous MSH2 mutation (Whiteside et al. 2002) and three patients carrying biallelic MSH6 mutations one of whom presented also with IgG2 deficiency (Ostergaard et al. 2005; Scott et al. 2007b).

Lynch syndrome features in CMMR-D patients

Microsatellite instability (MSI) in tumour tissue is a hallmark and diagnostic criterion of LS (Aaltonen et al. 1993; Boland et al. 1998; Peltomaki et al. 1993). It results from DNA mismatch repair deficiency attributable to the somatic biallelic inactivation of the involved MMR genes in the neoplastic cells. MSI has also been reported in all LS-associated tumours of CMMR-D patients so far analysed, i.e. 16 colorectal cancers and two duodenal carcinomas. MSI has also been tested in eight brain tumours, i.e. seven glioblastomas and one oligodendroglioma, but only three displayed an MSI phenotype (Hamilton et al. 1995; Hegde et al. 2005; Taylor et al. 1999). However, three of the patients with MS-stable brain tumours showed MSI in their LS-associated tumours or Wilms tumour, respectively (Agostini et al. 2005; Menko et al. 2004; Poley et al. 2007). Several suggestions have been made to explain the lack of MSI in the majority of the brain tumours. It is possible that MSI present in brain tumours of CMMR-D patients can not consistently be demonstrated by the microsatellite markers commonly used to test for MSI in colorectal tumours. However, MSI has been observed in glioblastoma tissues with biallelic (germ line and somatic) MLH1 and MSH2 mutations (Leung et al. 1998) as well as in approximately 27% of high-grade paediatric astrocytomas (Alonso et al. 2001). Alternatively, it has been suggested that MMR-gene deficiency could lead to brain tumour development by a mechanism distinct from a defect in the repair of postreplicative mismatches, such as resistance to apoptosis normally conveyed by MMR proteins (Bougeard et al. 2003; Poley et al. 2007). However, this hypothesis awaits further experimental evidence, as well. Whatever reasons account for the apparent differences in MSI-status in brain versus LS-related tumours, MSI seems to be a diagnostic tool with limited sensitivity to identify those patients with CMMR-D syndrome among the very young brain tumour patients. Immunohistochemistry showed lack of protein expression in all tested tumours (and surrounding normal tissue) from patients carrying truncating mutations (see Table 1). Hence, it might be a more suitable screening tool also in brain tumours.

As 50% of MMR activity confers normal levels of DNA repair, normal tissues from LS patients do not display an MSI phenotype. In CMMR-D patients, however, the constitutional lack of MMR could cause some extent of MSI even in normal tissues. Nine of 15 patients analysed showed MSI in their non-neoplastic tissues (see Table 1). It is of note that all these nine patients were analysed by small-pool PCR, i.e. the respective DNA samples were extensively diluted to approximately 0–3 genome equivalents per PCR-reaction (described by Parsons et al. 1995), while none of the six cases studied by conventional PCR essays displayed MSI in non-neoplastic tissue. Small-pool PCR may be more suitable for detection of MSI in normal tissues of CMMR-D patients than conventional PCR, as a particular microsatellite is not necessarily contracted or expanded likewise in every cell of a patient and instability in singular cells could be obscured if too many genomes are analysed at once. Support for this notion comes from two reports of patients who were MS-stable in their healthy tissues when analysed by conventional PCR, but displayed an MSI phenotype on reanalysis by small-pool PCR (Agostini et al. 2005; Felton et al. 2007a; Whiteside et al. 2002). Therefore, Felton et al. (2007a) suggested to develop standardised procedures for the detection of MSI in CMMR-D individuals according to the National Cancer Institute panel used for analysis of MSI in LS patients (Boland et al. 1998).

Preponderance of biallelic PMS2 mutations in CMMR-D syndrome

Roughly, half of the CMMR-D patients, i.e. 43 patients from 24 families, carried biallelic PMS2 mutations. This preponderance of PMS2 mutations may be in part explained by the presence of a PMS2 founder mutation that is responsible for CMMR-D syndrome in at least ten patients from four consanguineous Pakistani families living in the United Kingdom (De Vos et al. 2006). Nevertheless, if these patients are excluded mutant PMS2 alleles constitute the largest group of CMMR-D syndrome-causing biallelic mutations. One reason for the high prevalence of biallelic PMS2 mutations among CMMR-D patients may also be that CMMR-D patients due to PMS2 inactivation are more likely recognised as patients suffering from this syndrome than patients with biallelic mutations in the other MMR genes. Patients with MLH1 and MSH2 mutations often die at very young age from their first haematological malignancy or brain tumour, at least 9/20 (45%) patients (Bougeard et al. 2003; Poley et al. 2007; Ricciardone et al. 1999; Scott et al. 2007a; Vilkki et al. 2001; Wang et al. 1999). Since tumours of these organs—albeit other types—can be found also in NF1 patients, this may further hamper proper diagnosis of CMMR-D syndrome. In PMS2 mutation carriers the prevalence of second—often LS-associated—tumours is substantially higher than in MLH1/MSH2 deficient patients [17/42 (40%) vs. 3/20 (15%)]. Hence, the proper diagnosis is greatly facilitated in PMS2 when compared to MLH1/MSH2 mutation carriers. So far none of the patients carrying biallelic PMS2 mutations had a clear family history of LS, although some of their families showed an increased tumour incidence. This reflects the reported low penetrance of heterozygous PMS2 mutations and their little role for LS. However, a recent study of Truninger et al. (2005) shows that among unselected CRC patients heterozygous PMS2 mutations are as frequent as MLH1 and MSH2 mutations and account for a substantial fraction of microsatellite instable tumours. Therefore, the population frequency of heterozygous PMS2 mutations is likely to be higher than that of MLH1 and MSH2 mutations which cause LS with a high penetrance. This might be another factor worth to be considered as explanation for the higher prevalence of biallelic PMS2 mutation carriers. Another speculative hypothesis was brought forward in the context of Fanconi anaemia sybtype-D1 (FA-D1), a childhood cancer syndrome that shows overlap with CMMR-D syndrome and is caused by bi-parentally inherited mutations in the BRCA2 gene (Hirsch et al. 2004; Reid et al. 2005). There are data suggesting that at least some of the BRCA2 mutations occurring in breast cancer patients are not compatible with life in the homozygous state (reviewed by Rahman and Scott 2007). Likewise, it is possible that certain MSH2 and MLH1 mutations are not viable in a homozygous state, whereas this is less likely the case for PMS2 and, hence, biallelic PMS2 mutations are more prevalent in CMMR-D syndrome families. Evidence of differences in the MSH2 and MLH1 mutation spectrum in monoallelic and biallelic mutation carriers would render support for this theoretical possibility. However, so far the number of CMMR-D syndrome families carrying biallelic MLH1 and MSH2 mutations is too small to draw any such conclusions.

In any case, the high number of pathogenic PMS2 alleles in CMMR-D patients demands for robust, reliable and comprehensive PMS2 mutation analysis. However, sequence transfer—for simplicity referred to as gene conversion—between the PMS2 gene and a transcribed pseudogene, termed PMS2CL, severely compromises mutation analysis particularly in the 3′-half of the gene (Hayward et al. 2007). The process of gene conversion may be a mutational mechanism underlying PMS2 inactivation (Auclair et al. 2007), but it also creates functional PMS2 alleles containing “pseudogene-specific” variants at the 3′-end of the gene (Hayward et al. 2007; Etzler et al. 2008). Hence, the reference sequences of PMS2 and PMS2CL cannot be fully relied upon to distinguish the functional gene from its pseudogene. Different approaches have been developed to circumvent the resulting risks of allele drop-out as well as pseudogene coamplification. Long-range PCR approaches applicable to genomic DNA (Clendenning et al. 2006) largely overcome these problems, but are still not fully reliable for PMS2 exons 13–15 (Clendenning and de la Chapelle 2007). We developed an assay that is based on the selective amplification of PMS2 transcripts by long-range RT-PCR and subsequent sequencing of the generated RT-PCR-products. This RNA-based approach effectively circumvents the difficulties and pitfalls of gDNA-based mutation analysis in all PMS2 exons (Etzler et al. 2008).

Identifying the underlying genetic alteration in patients with CMMR-D syndrome has implications also for the wider family, because it facilitates identification of heterozygous individuals who are at risk for LS-related tumours. LS surveillance programs including regular colonoscopy and screening for endometrial as well as other LS-associated cancers occurring in the family are highly recommended to all adult heterozygous mutation carriers (Vasen et al. 2007). However, the risk for colon cancer and other LS-related tumours may be lower in heterozygous carriers of hypomorphic mutations identified only on the basis of a child homozygous or compound heterozygous for the mutation. In particular, the cancer risk of heterozygous PMS2 mutation carriers may be reduced when compared to individuals with MLH1 or MSH2 mutations. Therefore, large studies are needed to evaluate the tumour spectrum as well as the cost and emotional stress versus benefit ratio of the current surveillance protocols for heterozygous PMS2 mutation carriers and possibly adjust the age of onset and the frequency of the colonoscopies to the actual risk of this group. Currently, Senter et al. (2008) suggest offering PMS2 mutation carriers an intermediate screening regiment as was recommended by Lindor et al. (2006) for individuals with MSH6 mutations.

Have we so far seen only the tip of an iceberg?

The vast majority of the so far reported patients with CMMR-D are individual cases who have come to the attention of the physicians due to multiple malignancies in the patient or among siblings, the occurrence in a consanguineous family with a clear history of LS or due to the occurrence of a LS-associated tumour at very young age. Only very few studies have systematically applied selection criteria to search for patients with CMMR-D syndrome. De Vos et al. (2006) ascertained families through paediatric oncology services in three UK cities. The selection criteria applied were a history of parental consanguinity, early-onset cancer and one or more of the following characteristics: affected sibling, multiple tumour types and CLS. By these “stringent” criteria six kindreds of Pakistani origin (Pakistanis are the largest consanguineous social group in the UK) with 13 patients carrying homozygous PMS2 mutations were identified. Poley et al. (2007) queried their institution’s data base holding 2,230 paediatric oncology and haematology patients to identify 15 patients aged <16 years who had been diagnosed with multiple malignancies. The patients’ tumours were screened by MSI analysis and/or immunohistochemistry of MMR proteins for possible MMR defects. Applying these criteria the authors identified a patient compound heterozygous for MLH1 mutations and a very likely CMMR-D syndrome patient in whom mutation analysis was not possible, but one tumour showed loss of MSH6 protein. Although these reports suggest a low frequency of CMMR-D syndrome among paediatric cancer patients, there are also indications that the incidence of CMMR-D syndrome may be underestimated. A recent epidemiological study performed in Southern Sweden shows that the risk of childhood tumours in LS families is substantially increased (OR 29%) compared to control families (Magnusson et al. 2008). Of note, in none of the families with childhood cancers consanguineous marriages were apparent and none of the families showed evidence for LS in both the lines of the family. This finding may suggest that the possible CMMR-D patients in these families may be compound heterozygous for a highly penetrant MLH1, MSH2 or MSH6 allele causing LS in one branch of the family and a hypomorphic allele with reduced penetrance in the heterozygous state in the other branch of the family. None of these LS families carried a PMS2 mutation. Further support for a high number of CMMR-D patients among LS families comes from Trimbath et al. (2001) who analysed 23 reports of families with CLS and CRC. They found a decreased mean age of disease onset (i.e. 31.9 years) in these families when compared to LS families (44 years). Brain tumours, mainly glioblastomas but also medulloblastoma and others, were found in 47.5% (mean age of onset 16.5 years) of these families and lymphomas and leukaemia in 33.3% of the families. Many of the families have been published as Turcot syndrome families. Since recessive forms of Turcot have been shown to be caused by biallelic PMS2 mutations at least in a subgroup of patients not caused by adenomatous polyposis coli (APC) gene mutations (De Rosa et al. 2000; Hamilton et al. 1995), it is fair to retrospectively speculate that a number of these cases have to be considered CMMR-D patients.

Taken together, further careful search for CMMR-D syndrome among childhood cancer patients is needed to answer our question of whether CMMR-D syndrome is the underlying pathomechanism in only a small minority of childhood malignancies or whether we have so far seen only the tip of an iceberg.

Concluding remarks

In the past, the overlap with NF1 may have delayed or even prevented identification of the underlying genetic cause in patients suffering from CMMR-D syndrome. Now, almost 10 years after the first reports delineating the syndrome, the knowledge on the clinical presentation, the tumour spectrum and their pathological features has substantially increased. In view of the present knowledge, we suggest to suspect a biallelic mutation in one of the MMR genes MLH1, MSH2, MSH6 or PMS2 as the underlying disease cause in children and adolescents with malignancies other than clearly NF1-associated ones if they also present with one or more of the following: (atypical) CLS, consanguineous parents, a history of LS at least in one branch of the family, a second non-NF1 associated malignancy, a sibling with a childhood cancer. Extending the suspicion beyond patients with haematological, brain and LS-associated neoplasias and including also patients suffering from other childhood tumours such as Wilms tumour, rhabdomyosarcoma and neuroblastoma might elucidate the full spectrum of malignancies associated with CMMR-D syndrome.

As for LS, screening strategies for CMMR-D syndrome among paediatric cancer patients may include immunohistochemical and MSI analysis of the tumour, but also non-neoplastic tissue. MSI testing widely used as screening method in LS-associated tumours may have limitations for brain tumours and constitutive tissue in CMMR-D patients. Immunohistochemical expression analysis which has been shown to work well in colon carcinomas, brain and other solid tumours in CMMR-D patients confers the advantage to also guide subsequent mutation analysis in the four MMR genes.

With (RNA-based) long-range PCR strategies for PMS2 mutation analysis it is now possible to identify the underlying genetic alteration in CMMR-D syndrome patients in all four MMR genes. Determination of the causative mutation is important, because it facilitates identification and surveillance of heterozygous and homozygous individuals in the wider family, counselling of the 25% recurrence risk in siblings and allows for informed decision making about prenatal or preimplantation genetic diagnosis. Currently, LS surveillance programs are highly recommended to all adult heterozygous mutation carriers. The surveillance of homozygous children with a high risk for a wide spectrum of malignancy is difficult and currently there exist no recommendations in this respect. However, it is our hope that further research will delineate sensible and pragmatic approaches for surveillance and treatment of patients suffering from CMMR-D syndrome.

Acknowledgments

We thank Dr. Ludwine Messiaen for her helpful comments on the manuscript.

Copyright information

© Springer-Verlag 2008