Constitutional Mismatch Repair Deficiency (CMMRD)

Autosomal recessive cancer predisposition syndrome caused by biallelic pathogenic variants in one of the DNA mismatch repair genes (MLH1, MSH2, MSH6, PMS2), with repair deficiency present in all cells from the zygote, brain, haematological and gastrointestinal tumours in childhood and adolescence, cutaneous manifestations resembling neurofibromatosis type 1 and response to immune checkpoint inhibitors.

Literature and regulatory information updated: 7 October 2026.

0. Initial Reference Table

1. Introduction

Constitutional mismatch repair deficiency (CMMRD) is the recessive, biallelic form of germline alterations of DNA mismatch repair (MMR, mismatch repair) genes. Heterozygosity for a pathogenic variant in MLH1, MSH2, MSH6 or PMS2 defines Lynch syndrome, an autosomal dominant adult-onset predisposition; in that setting, repair deficiency arises only in tumour cells following somatic inactivation of the second allele [16, 17]. In CMMRD both alleles are inactivated in the germline, so repair deficiency exists in all tissues from conception, and the resulting phenotype is not a more severe Lynch syndrome but a distinct entity, with tumours in the first and second decades, cutaneous manifestations resembling neurofibromatosis type 1 and its own tumour spectrum [4, 12, 16]. The Oxford Desk Reference describes it as a severe DNA repair disorder in which survival into the second decade of life is exceptional and children develop colorectal adenomatous polyposis, primary brain tumours, leukaemia and café-au-lait macules [4].

The history of the entity predates its molecular definition. In 1959, Turcot, Després and St. Pierre published two cases of malignant central nervous system tumours associated with familial colonic polyposis [28]. The OMIM summary describes a brother with spinal cord medulloblastoma and colorectal adenocarcinomas and a sister with glioblastoma multiforme and pituitary adenoma, children of third cousins, leading to a proposal of autosomal recessive inheritance; in 1969, Baughman and colleagues described a sibship with recessively transmitted glioma and colonic polyposis, which they termed “glioma–polyposis syndrome”, and in 1992 Tops and colleagues showed haplotypes of the 5q21-q22 region incompatible with an APC-allelic cause in a sibship with Turcot syndrome [5]. According to the European C4CMMRD consortium review, these two siblings, with numerous colorectal adenomatous polyps, colorectal carcinoma and malignant brain tumours, are retrospectively, although without molecular confirmation, the first described CMMRD cases; in subsequent decades, patients with CMMRD and patients with familial adenomatous polyposis and brain tumours were grouped under the same name, “Turcot syndrome” [1]. In 1995, Hamilton and colleagues studied 14 families from Turcot registries and the original family: in 10 they found germline APC variants, with predominant medulloblastoma (11 of 14 patients, 79%) and a relative risk of cerebellar medulloblastoma 92 times that of the general population (95% CI 29–269) in familial adenomatous polyposis, whereas in the remaining four the brain tumour was glioblastoma multiforme, with replication errors characteristic of repair deficiency in the Turcot family and three others and germline hMLH1 or hPMS2 variants in two [29]. According to OMIM, the original family was analysed using autopsy slides of the glioblastoma and a rectal adenoma, with no DNA available for germline study; the MLH1 variant was found in family 14 (glioblastoma and colorectal cancer), and in family 12 (two siblings with colonic adenomas, glioblastoma, café-au-lait macules, rectal non-Hodgkin lymphoma at age 17 and rectal carcinoma at age 11), a truncating PMS2 variant was subsequently complemented by a second variant identified by De Vos and colleagues in 2004, demonstrating recessive inheritance [5, 11]. “Turcot” thus came to encompass two distinct aetiologies and is now considered historical: GeneReviews recommends discontinuing the term, and Orphanet transferred “Turcot without polyposis” and “Turcot with polyposis” subtypes to Lynch syndrome and familial adenomatous polyposis, respectively [6, 7, 17, 30].

The first molecular descriptions of CMMRD were published simultaneously in 1999. Ricciardone and colleagues identified three children with very early haematological malignancies, two with NF1 signs, in a Turkish family with hereditary non-polyposis colorectal cancer, and demonstrated homozygosity for a MLH1 nonsense variant [31]; Wang and colleagues, a French group, described two sisters born to parents heterozygous for an hMLH1 variant in a North African family with Lynch syndrome, with clinical signs of NF1 and early cancers (fatal non-Hodgkin lymphoma at age 2 in one; acute myeloid leukaemia at age 6 and medulloblastoma one year later in the other), and proposed NF1 as a preferential target of mutations accumulated during embryogenesis [18, 32]. In 2008, Wimmer and Etzler assembled 78 patients from 46 families and proposed the term “constitutional mismatch repair deficiency syndrome” to designate the underlying mechanism [1, 33], and suggested that prevalence was underestimated [5, 33]. In 2005, Ostergaard and colleagues, describing two compound-heterozygous MSH6 siblings with a clinical NF1 diagnosis, early central nervous system tumours and T-cell lymphoma, had assembled 27 published individuals with biallelic variants and proposed considering this aetiology in children with multiple café-au-lait macules and central nervous system tumours, haematological malignancies or early gastrointestinal malignancies; according to OMIM, these authors and Trimbath and colleagues considered the original Turcot definition too restrictive [5, 34].

International organisation of CMMRD knowledge depended on two consortia. The International Replication Repair Deficiency Consortium (IRRDC) was established in Toronto in 2007 and enrolled 160 families between 2007 and 2022; the European Care for CMMRD consortium (C4CMMRD) was founded in 2013, with a Paris-based registry containing more than 100 patients [18]. In 2014, C4CMMRD published a scoring system for testing indications and a surveillance protocol [1, 35]; surveillance recommendations from the American Association for Cancer Research (AACR) paediatric group and US Multi-Society Task Force on Colorectal Cancer followed in 2017 [2, 36]; international consensus group diagnostic criteria were published in 2022 [20]; and in 2024 comprehensive guidelines from the European Reference Network on Genetic Tumour Risk Syndromes (ERN GENTURIS), together with C4CMMRD, appeared with 82 recommendations, alongside the AACR update [15, 16]. Since the first descriptions, more than 300 patients have been published [18]. Two observations have transformed prognosis this decade: demonstration that systematic surveillance is associated with greater survival, and response of hypermutated tumours to immune checkpoint inhibitors, first described in two siblings with recurrent glioblastoma treated with nivolumab in 2016 [37, 38].

The clinical relevance of CMMRD exceeds its rarity. Diagnosis changes treatment of the first tumour (for example, temozolomide is no longer indicated in high-grade gliomas and immunotherapy becomes a consideration), requires surveillance for metachronous tumours and has consequences for the entire family, because parents and many relatives carry Lynch syndrome [16]. CMMRD is also a differential diagnosis of neurofibromatosis type 1 in children without an identifiable NF1 or SPRED1 variant, and of paediatric T-lymphoblastic lymphoma and high-grade glioma [39–41].

2. Molecular Genetics and Pathophysiology

2.1. The mismatch repair system and its constitutional failure

DNA replication is performed by polymerases whose fidelity depends on two successive controls: proofreading (proofreading) activity of the exonuclease domains of polymerases δ and ε, encoded by POLD1 and POLE, and post-replicative mismatch repair by MMR proteins [15, 16]. Base incorporation errors produce single-nucleotide variants, and polymerase slippage causes insertions and deletions concentrated in repetitive sequences, the microsatellites [15]. In the MMR system, the MutSα heterodimer (MSH2-MSH6) recognises mismatches and small insertion–deletion loops, the MutSβ heterodimer (MSH2-MSH3) recognises larger loops, and the MutLα heterodimer (MLH1-PMS2) is recruited by the MutS complex and activates PMS2 endonuclease activity, which cuts the newly synthesised strand for excision by exonuclease EXO1 [12, 17]. This architecture explains immunohistochemistry (IHC) patterns: MSH2 loss entails MSH6 loss, and MLH1 loss entails PMS2 loss, whereas isolated MSH6 or PMS2 loss points to the corresponding gene [4, 12]. As MSH6 cannot be replaced by MSH3 but MutSβ-dependent repair persists, MSH6-associated deficiency is partial and the microsatellite instability signal weaker, with direct consequences for diagnostic tests [12, 42, 43].

In Lynch syndrome, inactivation of the second allele occurs somatically during life in a cell of a target organ; in CMMRD both alleles are inactivated from the zygote and all cells in the body are MMR-deficient [16, 18]. This results in a high constitutional mutation rate, proposed as the driver of tumorigenesis in several organs [16]. The European group's 2026 review hypothesises that the early spectrum of T-cell lymphomas and high-grade gliomas results from convergence of systemic repair deficiency and intense cell division during development in tissues with high proliferation in early life, such as the thymus and brain with its gliogenic progenitors, unlike Lynch syndrome, in which the second hit occurs when the thymus has already involuted and gliogenesis has slowed [18]. Constitutional deficiency is not uniform across tissues: in the low-pass genomic instability characterisation (LOGIC) assay, repair deficiency indices (“MMRDness”) in normal cells were highest in the digestive tract, intermediate in blood and lowest in brain, increased over time in the same individual and were associated with earlier age at first cancer [44].

2.2. Tumour biology: hypermutation, ultrahypermutation and signatures

CMMRD tumours are intrinsically MMR-deficient, with high tumour mutational burden (TMB, tumour mutational burden) and microsatellite instability (MSI); hypermutation is usually defined as TMB ≥10 mutations per megabase [16]. Particularly in brain tumours, but also others, somatic acquisition of missense variants in POLE or POLD1 exonuclease domains is common, adding proofreading deficiency to repair deficiency and producing ultrahypermutated tumours, with TMB ≥100 mutations per megabase, specific mutational signatures and MSI undetected by classical fragment-length analysis panels but detected by more sensitive methods developed for constitutional DNA [15, 16]. In the IRRDC series treated with PD-1 inhibitors, median TMB was 233.8 mutations/Mb (range 3.4–912), tumours with MMR deficiency alone had a median of 15.8 mutations/Mb and those with combined MMR and proofreading deficiency 391.4 mutations/Mb; 67.7% of tumours in patients with CMMRD had combined deficiency, whereas no tumour in patients with Lynch syndrome did; COSMIC signature 6 was associated with Lynch syndrome and MMR-only deficient tumours, and signatures 10 and 14 with combined-deficiency tumours [21]. Extensive IRRDC genomic analyses showed that all CMMRD tumours have high mutation and microsatellite indel burdens and pathognomonic mutational signatures; central nervous system tumours had very high TMB but relatively lower “MMRDness”, haematological tumours the opposite and gastrointestinal tumours high values for both parameters [18, 19]. Enrichment in deleterious TP53, RAS-MAPK pathway and ATRX variants was described; the authors suggest that ultrahypermutation is viable only in cells that have lost TP53 [18].

Microsatellite instability strictly defined also distinguishes mechanisms: microsatellite indel signature analysis showed that MMR-deficiency signatures are dominated by multibase losses, whereas those associated with mutant polymerases contain mainly single-base gains, and that these signatures enable detection of repair deficiency in normal cells and prediction of immunotherapy response [45]. This biology has two therapeutic consequences: hypermutation generates immunogenic tumour neoantigens that render tumours sensitive to immune checkpoint inhibitors, and MMR deficiency confers resistance to drugs whose mechanism depends on a functional MMR system, particularly temozolomide [16, 46]. In primary MMR-deficient high-grade gliomas (priMMRD-HGG), an analysis of 162 tumours published in 2026 identified three subgroups defined by secondary driver mutations in replicative polymerases or IDH1: polymerase-mutant tumours are ultrahypermutated, with a “hot” immune microenvironment and immunotherapy response, whereas IDH1-mutant tumours are immunologically “cold” and resistant [47].

2.3. Non-neoplastic manifestations: pigmentation, NF1 and immunity

Café-au-lait macules and other NF1-like signs are the most common non-neoplastic manifestation. As early as 1999, NF1 was proposed as a preferential target of mutagenesis during embryogenesis [32]. Postzygotic mosaicism for pathogenic NF1 variants has been identified in at least three patients with CMMRD: a girl clinically diagnosed with sporadic NF1, with a thigh plexiform neurofibroma and freckling, developed glioblastoma with PMS2 loss in tumour and normal cells and ultrahypermutation; she had two PMS2 variants and an NF1 variant with allele frequencies of 20%, 3% and 9% in blood, skin and saliva; in a French cohort, 2 of 22 patients with CMMRD had NF1 mosaicism in blood DNA [48]. These data support NF1-like manifestations in some patients resulting from early NF1 mutations induced by repair deficiency [16, 18]. The same interpretation had been proposed by Ostergaard and colleagues in 2005 and has experimental support summarised in OMIM: Wang and colleagues (2003) found somatic NF1 alterations in 5 of 10 tumour cell lines with microsatellite instability and none of 5 repair-proficient lines, as well as in two primary tumours with instability [5, 10].

The MMR system also participates in immunoglobulin class-switch recombination and somatic hypermutation, necessary for B-cell maturation, so deficiency may cause IgG2/IgG4 subclass deficiency, IgA deficiency or, rarely, more severe phenotypes [16]. This knowledge is based on a 2008 study in three patients homozygous for deleterious PMS2 variants, describing a new form of intrinsic B-cell class-switch recombination deficiency, partial in vivo and marked in vitro, with deficient double-strand breaks in switch regions and abnormal switch junctions, suggesting a role for PMS2 in generating these breaks [49]. In the index case summarised in OMIM, a 12-year-old Turkish girl with multiple café-au-lait macules and homozygosity for a four-exon deletion with an 18 bp insertion in PMS2, recurrent infections began at age 1, immunodeficiency was diagnosed at age 9 and colorectal adenocarcinoma arose one year later; OMIM also records IgA deficiency in the two compound-heterozygous MSH6 siblings described by Ostergaard and colleagues [10, 11]. In a cohort of 15 unrelated patients (PMS2 n = 8, MSH6 n = 5, MLH1 n = 2), none showed clinical signs of primary immunodeficiency; IgM was increased in 3 of 12, decreased in 3 of 12 and normal in 6 of 12, and class-switched memory B cells were reduced in 5 of 12, with a lower frequency of somatic hypermutation in B-cell receptor rearrangements [50]. The AACR update states that clinically significant immunodeficiency and autoimmunity, previously described in some PMS2 cases, were not more frequent in the IRRDC cohort than in the general population [15].

2.4. Genes, distribution and inheritance

CMMRD follows autosomal recessive inheritance: patients are homozygous or compound heterozygous in the same MMR gene, and approximately half are homozygous for a pathogenic variant [16]. More than 60% of cases have biallelic PMS2 variants, more than 20% MSH6 variants and fewer than 20% MLH1 or MSH2 variants; this distribution reflects population heterozygote frequencies, with PMS2 and MSH6 approximately 2.5 to 4 times more prevalent than MLH1 and MSH2 [16]. The European guidelines cite a population PMS2 carrier frequency of 1 in 714 from first-degree relatives in the Colon Cancer Family Registry; GeneGuidance, based on the same study by Win and colleagues, gives 1 in 279 for Lynch syndrome overall, 1 in 1,946 for MLH1, 1 in 2,841 for MSH2 and 1 in 758 for MSH6 [12, 16]. As heterozygous penetrance in MSH6 and particularly PMS2 is low, families of patients with CMMRD frequently lack a history of Lynch-spectrum cancers: the Oxford Desk Reference notes that this history is often absent in parents and kindreds of children with PMS2-related CMMRD [4, 16]. In the first international consortium series (Bakry 2014; 23 children from 14 families), adult Lynch-spectrum tumours were observed in only 2 of 14 families, compared with childhood CMMRD-related tumours in all (p = 0.0007) [51]; according to OMIM, only 1 of 28 parents had a cancer history, a mother who died aged 25 with an unspecified brain tumour and who, given multigenerational consanguinity, probably also had CMMRD [5]. The AACR update adds that parents rarely develop cancer young enough to have been diagnosed with Lynch syndrome when their child reaches paediatric oncology [15].

Deletions of the 3′ end of EPCAM, located upstream of MSH2, cause Lynch syndrome through secondary MSH2 promoter methylation in epithelial tissues [17]. In homozygosity, these deletions produce tissue-specific repair deficiency: in a patient with a homozygous EPCAM deletion, induced pluripotent stem cell-derived colon organoids showed MSH2 promoter hypermethylation, MSH2 loss, increased mutational burden and MMR-deficiency signatures, whereas brain organoids, blood and fibroblasts retained functional repair [14]. Clinically, patients develop multiple gastrointestinal cancers without other-organ involvement, and the AACR recommends annual upper and lower gastrointestinal surveillance from diagnosis [15, 18]. Biallelic MSH3 variants, however, cause adult-onset recessive adenomatous polyposis, and biallelic truncating MLH3 variants (Finnish founder c.3563C>G, p.Ser1188*) cause polyposis and other adult cancers; neither has sufficient evidence to recommend CMMRD-like surveillance [15, 30].

2.5. Genotype–phenotype correlations

In the IRRDC cohort (201 patients from more than 50 countries, enrolled between 2007 and 2022), MLH1 or MSH2 variants were associated with earlier cancer onset and poorer survival than PMS2 or MSH6 variants: overall survival at age 15 was 63% (95% CI 55–73) for PMS2, 49% (35–68) for MSH6, 19% (6–66) for MLH1 and 0% for MSH2; within the same gene, frameshift or truncating variants were associated with earlier cancers and poorer outcomes than missense variants, and these differences persisted despite overall survival improvement with surveillance and immunotherapy [19]. The 2014 C4CMMRD review had already suggested an overall more severe phenotype in biallelic MLH1/MSH2 carriers and noted that the proportion with more than one malignancy was lowest in biallelic MLH1/MSH2 carriers and highest in PMS2 carriers, interpreting this as a greater likelihood of surviving the first tumour [1]. A 2025 systematic review of PMS2-related CMMRD divided published cases into onset before and after age 10 and observed predominantly central nervous system tumours in the early group and gastrointestinal tumours in the later group, identifying six PMS2 variants associated with one or other pattern, conclusions the authors consider to require prospective validation [52].

Hypomorphic variants explain attenuated forms. In seven Inuit families from northern Quebec, PMS2 variant NM_000535.7:c.2002A>G creates a splice donor site competing with the authentic one; in homozygosity, full-length protein is reduced to almost undetectable levels, but median age at first cancer was 22 years in 13 homozygotes, versus 8 years in biallelic truncating-variant carriers, and residual full-length transcript expression was detected in normal tissues of homozygotes with cancers in their third decade [53]. The OMIM summary adds that the splicing defect leads to nonsense-mediated messenger RNA decay, with small amounts of full-length transcript and some residual functional protein; that brain tumours were significantly less frequent in c.2002A>G homozygotes (15%) than truncating-variant homozygotes (67%), although one homozygote developed a primitive neuroectodermal tumour at age 3; and that haplotype analysis dated the founder effect to the late 11th century [11]. These figures derive from 13 patients from seven families and comparison with a truncating-carrier series, so describe a group trend and cannot exclude paediatric tumours in a particular homozygote. The molecular effect of this variant involves loss of the last 5 nucleotides of exon 11 (r.2002_2006del) with premature termination codon p.Ile668*, coexisting with residual expression of full-length transcript encoding a functional protein with p.Ile668Val; Gallon and colleagues therefore represent it as p.[(Ile668*, Ile668Val)], and GeneReviews describes the same novel splice site with premature truncation [17, 54]. In 2024, two patients with first cancer at ages 27 and 38, initially investigated as Lynch syndrome, were described, homozygous for MSH6 NM_000179.3:c.3226C>T p.(Arg1076Cys) and MLH1 NM_000249.4:c.306G>A p.(Glu102=) (synonymous variant with partial exon 3 skipping); constitutional MSI detection confirmed CMMRD; the literature review assembled 13 cases with first cancer at age 25 or older, with 18 colorectal carcinomas among up to 27 cancer diagnoses, and the authors advocate individualised management of hypomorphic-variant carriers [54]. These cases do not meet age-based testing indication criteria, representing a known limitation [54].

The relationship between constitutional instability and phenotype is not uniform across assays. With LOGIC, higher “MMRDness” indices were associated with earlier first cancer (P = 2.2 × 10⁻⁵) [44]; with high-sensitivity MSI (hs-MSI), correlation with first tumour age was negative and moderate (r = −0.43) [55]; the constitutional MSI (cMSI) index based on 32 mononucleotide repeats varied with genotype (lower in MSH6 deficiency and with at least one missense variant, higher with biallelic truncating or copy-number variants), but did not correlate with first tumour age, leading the authors to conclude that it is not a useful risk predictor [43].

2.6. Illustrative normalised variants

The table assembles variants cited in this article, normalised to MANE Select transcripts using VariantValidator; where the original publication used another transcript version or notation, this is indicated in parentheses [13].

3. Clinical Spectrum and Manifestations by System

3.1. Overview and natural history

CMMRD is probably the cancer predisposition syndrome with the highest cancer risk, with tumours described from the first year of life; according to the review of published cases, the C4CMMRD database and the 2024 IRRDC cohort (with partial patient overlap), 80% to 90% develop their first tumour before age 18 and approximately half before age 10 [16]. In the IRRDC cohort (201 patients, 103 male, enrolled between June 2007 and September 2022, median follow-up 7.2 years), 339 cancers were recorded in 194 patients (97%), cumulative cancer incidence by age 18 was 90% (95% CI 80–99) and, in patients with more than one cancer, the median interval between diagnoses was 1.9 years (IQR 0.8–3.9); tumours arose in 15 organs [19]. In the earlier analysis from the same consortium (110 patients, 193 malignant tumours), median age at first cancer was 9.2 years (1.7–39.5), and 36 of 52 patients surviving their initial cancers (69%) developed metachronous cancers [37]. The AACR summarises risk as 90% by age 18 and almost all patients by age 40, with a new cancer on average every 2 years in survivors and a 25% risk of presentation with synchronous cancers [15]. These figures derive from diagnosed-patient cohorts, many after a first tumour, and the European guidelines warn that ascertainment bias may overestimate risks; they should also therefore not be extrapolated to hypomorphic-variant carriers [16, 54].

Tumour distribution by group and ages at diagnosis are consistent across the three main series, as summarised in the table.

Source: table 1 of the review by Vasen and colleagues (2026), compiling Wimmer 2014, Durno 2021 and Ercan 2024; the 2024 series includes the 2021 series [18].

The first international consortium series, published in 2014 by Bakry and colleagues, predates these cohorts and shows the same pattern on a smaller scale: 22 of 23 children with CMMRD from 14 families (96%) developed 40 tumours, of which 48% were cerebral, 32% gastrointestinal and 15% haematological, and 12 (30%) were resectable low-grade cancers [51]. The OMIM summary adds that, among central nervous system malignancies, 74% were high-grade gliomas, 16% low-grade gliomas and 10% medulloblastomas or primitive neuroectodermal tumours, and that 3 of the 4 children without a tumour at referral developed tumours during the study [5]. With 23 patients referred to a consortium, these proportions are descriptive rather than penetrance estimates; the 2024 IRRDC cohort is now the quantitative reference [19].

3.2. Central nervous system

Brain tumours, particularly high-grade gliomas, are the most common malignancy in CMMRD, with median diagnostic age approximately 9 years; they present with raised intracranial pressure (headache, nausea, vomiting, diplopia), grow rapidly and are frequently diagnosed at an advanced stage [18]. In the C4CMMRD registry, 49 of 87 patients developed 56 malignant brain tumours (50 high-grade gliomas and 6 embryonal tumours), with median age 9.2 years (1.1–40.6) and nine patients older than 18; multinucleated giant cells were observed in 16 of 21 reviewed gliomas, 27 patients had multiple malignancies, 40 of 41 had café-au-lait macules and 11 of 15 had developmental brain anomalies; overall survival was 30% at 3 years and 22% at 5 years after the first brain tumour, with 38 of 43 deaths (88%) related to the brain tumour [61]. In the 2024 IRRDC cohort, patients with central nervous system tumours had the poorest overall survival at 10 years after diagnosis (39%, 95% CI 30–52) [19]. The spectrum also includes medulloblastoma, ependymoma and other gliomas [15]. On magnetic resonance imaging, CMMRD-associated high-grade gliomas are predominantly hemispheric (91% versus 25% in NF1 and 54% without predisposition) and poorly defined [62]. An analysis of primary MMR-deficient, IDH-wildtype and H3-wildtype gliomas published in 2026 describes them as a distinct type rich in multinucleated giant cells, occurring in children, adolescents and older adults and enriched among patients with CMMRD and Lynch syndrome, with prolonged survival in tumours with TMB over 50 mutations/Mb and intact 9p region [63].

3.3. Haematopoietic system

Haematological malignancies arise mainly in childhood and are the first (sometimes only) malignancy in approximately one-third of patients; T-cell non-Hodgkin lymphomas, particularly mediastinal, are most common [64]. A collaborative C4CMMRD, IRRDC and European Intergroup for Childhood Non-Hodgkin Lymphoma (EICNHL) study analysed 74 patients with CMMRD-associated lymphoma registered before 2020: median age 9.4 years, previous malignancy in 36%, café-au-lait macules in 96%, consanguinity in 54%; initial subtypes were 53 T-lymphoblastic lymphomas, 4 B-lymphoblastic lymphomas and 17 mature B-cell lymphomas; 20 patients had multiple metachronous lymphomas [65]. In the complete Dutch cohort of 88 children with T-lymphoblastic lymphoma diagnosed between 2007 and 2020, 8 (9.1%) had CMMRD (PMS2 5, MSH6 3), all with multiple café-au-lait macules, all with a mediastinal mass, and 6 of 8 with previous or subsequent malignancies [40]. The haematological spectrum also includes B-cell, mixed-lineage and myeloid leukaemias and Hodgkin lymphoma [15]. Symptoms depend on subtype and location: cough and dyspnoea in mediastinal T-lymphoblastic lymphoma (with risk of superior vena cava syndrome), bowel obstruction in B-cell lymphoma, and anaemia, thrombocytopenia, neutropenia, lymphadenopathy, bone pain, fever and bleeding in acute lymphoblastic leukaemia [16, 18].

3.4. Digestive tract

Most patients develop gastrointestinal adenomas, frequently with high-grade dysplasia or oligopolyposis, in their second decade [16]. Colorectal adenomatous polyps arise at a median age of 9 years (described from age 6), with highly variable burden (a few to more than 100 polyps) and faster progression to carcinoma than in Lynch syndrome; colorectal carcinoma is diagnosed at a median of approximately 16 years, small-bowel adenomas at a median of 12 years and small-bowel carcinomas between 21 and 28 years [18]. The AACR considers polyposis almost universal, with distinct polyps in the upper and lower digestive tract detectable from age 6 [15]. In a series of 11 Israeli patients assessed by colonoscopy, upper endoscopy and small-bowel investigation, findings ranged from no polyps (two patients aged 6), few adenomas (ages 8 to 14), a polyposis-like phenotype (ages 13 to 16, two with high-grade dysplasia or cancer and one with atypical juvenile polyps) and colorectal carcinoma at ages 19 and 20 [66]. Small-bowel and gastric cancers have been diagnosed in patients as young as 9 [16]. In the IRRDC cohort, 11 of 19 patients with at least 3 years of digestive surveillance (58%) developed dysplastic polyps leading to prophylactic colectomy by age 13.2 (9.2–25), and cumulative probability of high-grade dysplasia in colorectal polyps was 81% over 8 years [37]. Gastrointestinal cancers had the best survival among major tumour groups, at 89% at 10 years (95% CI 81–97) [19].

3.5. Other solid tumours and young-adult tumours

Other tumours occur in 3.5% to 7% of cases, including neuroblastoma, Wilms tumour, ovarian neuroectodermal tumour, infantile myofibromatosis, rhabdomyosarcoma, osteosarcoma and other sarcomas, parotid mucoepidermoid carcinoma, retinoblastoma, breast cancer, ovarian yolk sac tumour, osteochondroma, pilomatrical neoplasms, basal cell carcinoma, sebaceous carcinoma and melanoma [18]. Rhabdomyosarcoma was documented in two consanguineous families, one with homozygosity for a PMS2 variant and another with isolated PMS2 loss in tumour and adjacent normal tissue [67]; according to OMIM, in the first family a boy had rhabdomyosarcoma at age 3 and colonic adenocarcinoma at age 8 (homozygous for the variant termed C73X, historical notation; the only single-nucleotide substitution converting codon 73 (TGT, cysteine) into a stop codon corresponds to NM_000535.7:c.219T>A p.(Cys73Ter), without confirmation at nucleotide level in the source), and in the second two siblings had multiple colonic adenocarcinomas and, respectively, anaplastic astrocytoma and undifferentiated sarcoma, without possible genetic study [11]. The authors concluded that rhabdomyosarcoma and possibly other embryonal tumours such as neuroblastoma and Wilms tumour belong to the syndrome spectrum and suggested family analysis and MMR expression assessment in children with embryonal tumours and NF1 signs [67]. With improved survival, young adults develop adult-type cancers (breast, prostate, genitourinary, pancreatic, melanoma) at younger-than-usual ages, and genitourinary cancers have been recorded from age 14 [15]. The updated IRRDC database reported 14 skin tumours (3 melanomas, 10 pilomatrical neoplasms and 1 sebaceous carcinoma) [18]. The 2014 review also recorded hepatic adenomas in three genetically confirmed patients [1].

3.6. Skin and neurofibromatosis type 1-like manifestations

Café-au-lait macules are the most prevalent non-neoplastic manifestation, alongside other hyperpigmented and hypopigmented changes [16]. In the IRRDC cohort, 117 of 126 patients with complete skin data (93%) had dermatological manifestations and 35 of 126 (28%) met clinical NF1 criteria [19]. In the review of 146 patients, 27 had other diagnostic NF1 signs in addition to macules: freckling in 14, neurofibromas in 8, Lisch nodules in 5, and tibial pseudarthrosis, sphenoid wing dysplasia and optic glioma in one patient each; several patients had fewer than six macules and many lacked the typical NF1 appearance, and some had a segmental distribution of signs [1]. Atypical macules had been noted previously: in the series of 13 patients homozygous for PMS2 variants from six families of Pakistani origin, the phenotype included café-au-lait macules and could be mistaken for NF1 [58]; according to OMIM, 10 of 13 had macules (no information for the other 3), described by the authors as irregularly bordered and slightly indistinct, unlike the sharply demarcated macules typical of NF1 [11]. In Bakry's 2014 series, all children had café-au-lait macules, and OMIM also mentions hypopigmentation and axillary freckling; MSH6 reports summarised in OMIM also rarely include Lisch nodules [5, 10, 51]. Cutaneous hypopigmentation, pilomatrixomas and multiple cerebral venous anomalies are more common in CMMRD than NF1, and cognitive and learning difficulties less common [15]. The comparison with NF1 is detailed in section 6.

3.7. Nervous system: non-neoplastic anomalies

Developmental venous anomalies (DVA, developmental venous anomalies, also called cerebral venous angiomas) are present in 73% to 100% of patients with CMMRD, versus 1.6% in the general paediatric population [16]. In a series of 10 patients from 3 families, all had two or more DVAs, two had non-treatment-induced cavernomas, none had symptomatic haemorrhage and tumours were not adjacent to DVAs [68]. In a series of 14 children, 12 (85%) had DVAs, 3 (21%) cavernomas and 10 (71%) non-specific subcortical T2 hyperintensities, three of which evolved into tumours [69]. Comparing 30 children with CMMRD (22 with high-grade glioma) with patients with Lynch syndrome, NF1 or no predisposition, all children with CMMRD had at least one DVA (versus 14%, 10% and 6%), 83% had multiple DVAs (versus one NF1 patient and none in the other groups) and 21% cavernomas (versus 0%); focal T2-FLAIR hyperintense areas resembling those in NF1 were more prevalent than in Lynch syndrome or without predisposition, and poorly demarcated subcortical areas were specific to CMMRD [62]. Agenesis of the corpus callosum, with or without grey matter heterotopia, was described in at least 4 of 60 patients (6.6%), far above the birth prevalence of these malformations in the general population [70]. The denominator comprises 57 patients with brain tumours published by then (who would have undergone brain imaging) plus the three new cases, and OMIM presents the figure as a prevalence that “may reach” 6.6%; it is therefore a proportion in published cases, rather than a systematic estimate [11, 70]. In the three patients of Baas and colleagues summarised in OMIM, corpus callosum agenesis was associated with interhemispheric cysts and periventricular heterotopias in two (one homozygous for MLH1 and one with PMS2), and in a third with PMS2 had been suspected from lateral ventricular dilatation on prenatal ultrasound; psychomotor development was normal in the two PMS2 patients and slightly delayed in the MLH1 patient [5, 11].

3.8. Immunity and autoimmunity

Immunoglobulin class-switch deficiency may manifest as reduced IgG2/IgG4 or IgA, with elevated IgM in young children, described as hyper-IgM-like in patients with PMS2 and MSH6 [1, 16]. Individual reports nevertheless include recurrent infections from the first year of life and class-switch recombination deficiency immunodeficiency in a homozygous PMS2 patient (section 2.3) [11, 49]. In the systematic cohort of 15 patients, none had clinical immunodeficiency signs [50]. Paediatric systemic lupus erythematosus, rare in the general population (prevalence below 1 in 25,000 children), was found in six of more than 200 published patients with CMMRD, leading to its inclusion as a one-point criterion [16, 71].

3.9. Initial gene-specific reports summarised in OMIM

The four OMIM entries organise CMMRD by gene and summarise the reports defining each subtype. These are descriptions of isolated families selected by severity and publication, illustrating phenotypic breadth but not allowing gene-specific frequency estimates; quantitative gene distribution comes from the cohorts in section 2.5 [5, 9–11].

Sources: OMIM #276300, #619096, #619097 and #619101 [5, 9–11]. The families of Hamilton (1995), De Vos (2006), Peron (2008), Giunti (2009), Kratz (2009), Baas (2013) and Li (2015), also summarised in OMIM, are addressed in sections 1, 2.3, 2.5, 3.5, 3.7, 5.2 and 9.2.

4. Clinical Recognition and Warning Signs

Recognising CMMRD requires assembling features that are individually non-specific: a particular paediatric cancer, NF1-like skin signs, brain imaging findings, family history of Lynch syndrome or consanguinity and tumour molecular features. The 2026 European review summarises suspicion as follows: consider CMMRD in children with high-grade glioma, lymphoblastic lymphoma, colorectal carcinoma or multiple colorectal adenomas (after excluding APC- or MUTYH-related polyposis), recognising that in principle any cancer in a child or young adult may be associated with the syndrome [18]. Diagnosis is frequently missed for three reasons: rarity, recessive inheritance and more than half of cases being due to PMS2, whose heterozygosity confers low cancer risk, so family history of Lynch-spectrum tumours is often negative and heterozygous parents are usually young and healthy when their child becomes ill [18]. In the IRRDC cohort, endogamy in minorities and closed communities explained high homozygosity even in countries with low consanguinity [19]; absence of reported consanguinity therefore does not exclude diagnosis [15].

The colorectal cancer chapter of the Oxford Desk Reference recalls that homozygosity for MMR genes should be considered in consanguineous families, with café-au-lait macules and depigmented lesions, brain tumours and gastrointestinal polyps, and records CMMRD as a differential diagnosis of café-au-lait macules and NF1, with a phenotype of macules, axillary freckling, mild generalised increased pigmentation, primitive neuroectodermal tumours, non-Hodgkin lymphoma and intestinal polyps [72–74]. Tumour clinical presentations require patient and family education about warning symptoms: dyspnoea and superior vena cava syndrome in mediastinal lymphomas, pancytopenia symptoms in leukaemias, neurological signs in brain tumours and gastrointestinal bleeding in colorectal tumours [16].

5. Diagnosis: Criteria, Tests and Algorithms

5.1. Whom to test

The 2024 ERN GENTURIS guidelines recommend offering CMMRD testing to: (1) patients with cancer scoring at least 3 points under revised C4CMMRD criteria; (2) patients younger than 18 with a tumour showing “paediatric-high” TMB, with or without a somatic POLE or POLD1 variant; (3) patients with cancer whose tumour shows loss of one or more MMR proteins in neoplastic and non-neoplastic cells (infiltrating lymphocytes, endothelial cells); (4) patients younger than 18 with a heterozygous (likely) pathogenic MMR gene variant on germline sequencing; and (5), following interdisciplinary discussion, cancer-free children with suspected sporadic NF1 or Legius syndrome, without an NF1/SPRED1 variant after comprehensive testing, with at least one additional feature (all strong recommendations) [16]. Recommendation 4 is justified because the second variant, particularly in PMS2, may escape exome or genome sequencing [16]. The 2026 review adds that patients with Lynch syndrome with carcinoma before age 18 should undergo testing to exclude a second variant unknown in the family [18].

Source: table 2 of the ERN GENTURIS guidelines [16], also reproduced in the 2026 review [18].

Changes from the 2014 score were deliberate: T-cell non-Hodgkin lymphoma became T-lymphoblastic lymphoma; “supratentorial PNET” ceased to exist as an entity and was absorbed into diffuse high-grade gliomas; multiple DVAs were included with 2 points and paediatric lupus with 1 point; and two or more pigmented macules changed from 2 to 1 point because Gustave Roussy's experience with the original criteria showed they were too inclusive, generating unnecessary anxiety and diagnostic effort [16]. The original 2014 score assigned 3 points to highly specific tumours, 2 to overrepresented malignancies and 1 to the remainder, and weighted pigmented macules, brain malformations, pilomatrixomas, second malignancy, Lynch-spectrum tumour in a relative and consanguinity at 1 to 2 points [1]. Guidelines also advise considering CMMRD in patients with genetically confirmed NF1 (including mosaicism) developing an atypical NF1 malignancy, such as diffuse high-grade glioma or T-lymphoblastic lymphoma [16].

For cancer-free children with suspected NF1 or Legius syndrome, the C4CMMRD strategy requires as prerequisites at least one diagnostic NF1 sign (including at least two suggestive café-au-lait macules), no NF1 or SPRED1 variant after high-sensitivity analysis (including methods for unusual splice variants) and no NF1 signs in either parent; plus at least one additional family feature (consanguinity, Lynch syndrome on one parental side, sibling with NF1 signs, sibling deceased with childhood malignancy, Lynch-spectrum carcinoma before age 60 in a first- or second-degree relative) or patient feature (atypical macules, multiple hypopigmentation, pilomatrixoma, corpus callosum agenesis, non-treatment-induced cavernoma, multiple DVAs) [16, 39]. Empirical justification for this preselection came from screening 735 children from Birmingham (Alabama, USA) and Innsbruck (Austria) laboratories with suspected sporadic NF1 without an NF1/SPRED1 variant: leucocyte MSI was positive in 7 and CMMRD was confirmed in 3 (0.41%; 95% CI 0.08–1.19%) [56]. A healthy 6-year-old child of consanguineous parents with more than six macules and freckling, found homozygous for a PMS2 variant, illustrated the benefits and risks of diagnosing CMMRD before the first cancer [75].

5.2. How to test: laboratory strategy

The aim of any strategy is a definitive diagnosis confirming or excluding CMMRD and identifying causal variants [16]. Whenever (pre-)malignant tissue is available, IHC of all four MMR proteins should assess expression in neoplastic and non-neoplastic cells; loss in both is the characteristic CMMRD marker, whereas loss limited to tumour cells indicates somatic deficiency and should guide Lynch syndrome investigation [15, 16]. This principle dates to the first international consortium series (Bakry 2014): four-protein IHC in 26 available tumours showed corresponding protein loss in all 17 tumours from patients with CMMRD, intratumoral normal-tissue staining was negative in agreement with genetic confirmation, and the authors concluded that tumour IHC was 100% sensitive and specific for corresponding gene deficiency, whereas MSI, studied in 28 tumours and matched normal tissues, was neither sensitive nor specific compared with IHC (p < 0.0001) [5, 51]. These statements should be read as historical data from a small series, without confidence intervals and with IHC serving simultaneously as test and reference; the MSI conclusion concerns classical tumour panels of that era rather than subsequently validated constitutional MSI assays (section 5.4). Current consensuses differ over IHC as an ancillary confirmatory test, accepted by the 2022 international group and excluded by the 2024 European guidelines (section 5.4) [16, 20]. The same authors proposed an algorithm for those younger than 18 with café-au-lait macules and consanguinity, or macules and a relative with CMMRD-related tumours: four-protein IHC in tumour or normal tissue followed, if expression is lost, by genetic testing to initiate surveillance and counsel the family [5]; this approach has been replaced by the testing indication criteria in section 5.1. IHC has limitations: poor fixation, missense variants retaining expression, partial loss and interobserver variability; a negative result in a suggestive clinical setting should be interpreted cautiously [15, 18]. In prospective screening of 127 paediatric high-grade gliomas in the HIT-HGG-2013 trial, MMR IHC identified all Lynch syndrome or CMMRD cases (5.5%), with 100% sensitivity and 96% specificity against germline sequencing, and also detected 2.3% of cases with exclusively somatic MMR alterations [76]. Classical fragment-analysis MSI panels are unreliable in ultrahypermutated tumours or constitutional DNA in these children and are not recommended for this purpose [15, 18]. The biological reason was described in gliomas: with a five-mononucleotide-marker panel, only 2 of 34 paediatric gliomas showed instability, both glioblastomas from patients with PMS2 or MLH1 variants and a compatible family history; instability was subtle, type A (allelic shifts of 1 to 2 bp), whereas a colorectal carcinoma from an affected sister had type B instability with larger shifts, leading the authors to propose type A MSI in paediatric gliomas as an indicator of this syndrome [11, 77]. Conversely, instability in more than 40% of markers was described in a glioblastoma from a patient with biallelic MSH6 [10], illustrating tumour-result variability.

The genetic testing laboratory should be able to analyse all four MMR gene transcripts and use methods circumventing PMS2–PMS2CL pseudogene homology, itself or in partnership [16]. For index cases, it should have access itself or in partnership to at least one validated ancillary test capable of confirming or excluding CMMRD when genetic testing is inconclusive: the wording is “should probably” for patients with (pre-)malignancy and “should” for those without malignancy, both with strong recommendation strength (recommendations 10 and 11) [16]. The AACR recommends multigene panels including all four genes and close mimics, and notes that rigorous PMS2 analysis may require long-range PCR available only in specialist laboratories [15]. If CMMRD is not confirmed in patients indicated by clinical criteria, TMB or a heterozygous variant, germline variants in POLE and POLD1 exonuclease domains should be sought (recommendation 13), because certain heterozygous POLE variants and digenic combinations of POLE/POLD1 with a heterozygous MMR variant produce a similar phenotype [16, 78].

5.3. PMS2, pseudogenes and other germline testing limitations

PMS2 is the most frequently involved gene and the most difficult to analyse. High homology with pseudogenes, particularly PMS2CL, and frequent sequence-exchange events between gene and pseudogene generate hybrid alleles complicating conventional sequencing and interpretation [1, 17]. GeneReviews recommends selecting a laboratory with specific gene expertise following American College of Medical Genetics and Genomics (ACMG) standards, mentions strategies such as long-range PCR, complementary DNA sequencing and specific next-generation sequencing approaches, and estimates that large rearrangements account for 20% to 50% of pathogenic PMS2 variants [17]. In a Dutch national study of 396 probands with suspected Lynch syndrome or CMMRD using improved DNA and RNA strategies, 52 different pathogenic PMS2 variants were found, explaining 121 patients with Lynch syndrome and 9 with CMMRD; 10% of isolated PMS2-loss cases had MLH1 variants, and 6% of carriers showed discordant IHC [79]. The functional assay of Shuen and colleagues was developed precisely because genetic diagnosis is confounded by 15 or more pseudogenes and variants of uncertain significance [80]. Other known limitations: constitutional MLH1 epimutation is undetectable by sequencing or deletion/duplication analysis; only EPCAM deletions including the last exon cause MSH2 silencing; and variants should be classified according to gene-specific ClinGen and InSiGHT expert panel specifications [16, 17].

A useful CMMRD germline report should state the gene, versioned transcript (MANE Select: MLH1 NM_000249.4, MSH2 NM_000251.3, MSH6 NM_000179.3, PMS2 NM_000535.7), complementary DNA and protein HGVS nomenclature, zygosity, ACMG/AMP classification according to ClinGen-InSiGHT specifications, coverage (sequence, copy number, PMS2-specific methods, transcript analysis) and limitations [13, 16, 17]. Two variants in the same gene support CMMRD only if in trans, requiring parental, offspring or other-relative studies, or direct phasing methods [15, 16]. A variant of uncertain significance does not become pathogenic by coexisting with another in trans; confirmation then depends on ancillary tests and, where applicable, transcript analysis [16].

5.4. Ancillary tests of constitutional repair deficiency

Microsatellite instability exists at low levels in non-neoplastic tissues of patients with CMMRD, and several assays have been developed to detect it in blood or other tissues. The table summarises tests recognised by the European guidelines and other assays with published validation.

The European guidelines require each test to be validated within the laboratory using at least eight patients with CMMRD of different genotypes (including one per gene), at least twenty adults older than 40 without cancer or an MMR variant, at least ten MMR heterozygotes and, if possible, POLE/POLD1 variant carriers [16]. According to the guidelines, all except gMSI achieved 100% sensitivity and specificity in validation cohorts (blinded hsMSI validation published in 2024 obtained 98.5% sensitivity and 100% specificity [55]), and a positive result from any, including gMSI, can confirm CMMRD when genetic testing is inconclusive [16]. Unlike the 2022 international criteria, European guidelines do not include IHC as an ancillary test, because it may generate false positives and false negatives, except in a deceased patient without available germline DNA [16]. The AACR estimates costs below 150 dollars per sample for genomic MSI-based functional assays, considers them useful for resolving ambiguity even in lower-resource settings and reports that one such assay, available through IRRDC, was undergoing accreditation [15]. European authors recommend constitutional MSI analysis at reference laboratories [18].

5.5. Diagnostic confirmation criteria

The 2024 European criteria (recommendation 12, moderate strength) consider CMMRD confirmed in an individual with a testing indication when: (i) two (likely) pathogenic variants are in trans in the same gene; or (ii) two variants are in trans in the same gene, one (likely) pathogenic or of uncertain significance and the other of uncertain significance, and at least one validated ancillary test is compatible; or (iii) there is a single (likely) pathogenic variant or variant of uncertain significance, transcript-analysis evidence of abnormal splicing unexplained by the variant or reduced wild-type allele expression, and a positive ancillary test; or (iv) no variant is identified, but there is a positive ancillary test and evidence of abnormal splicing or reduced expression in an MMR gene [16]. For some combinations (for example, a pathogenic with a likely pathogenic variant in a patient with high TMB), confirmation requires exclusion of germline POLE/POLD1 variants; in a deceased patient with cancer without germline DNA/RNA, loss of one or more MMR proteins in neoplastic and non-neoplastic cells with an appropriate positive control is considered confirmatory (recommendation 14) [16].

The 2022 international consensus group (IRRDC, C4CMMRD and AACR) proposed seven criteria: four for definitive and three for likely diagnosis, all indicating surveillance [20]. The table summarises them as reproduced by the AACR in 2024.

CMMRD hallmark cancer: glioma or embryonal central nervous system tumour before age 25, haematological cancer (excluding Hodgkin lymphoma) before age 18, gastrointestinal adenocarcinoma before age 25 or more than 10 gastrointestinal adenomatous polyps before age 18 (other polyposis syndromes excluded). Source: AACR 2024 table 1, reproduced from Aronson and colleagues [15, 20].

The approaches differ on a matter of principle: European guidelines do not distinguish definitive from likely diagnosis and insist on obtaining conclusive evidence given the consequences of CMMRD diagnosis, whereas the AACR recommends using the 2022 criteria and applying surveillance to both definitive and likely cases, even without confirmed genetic diagnosis [15, 16, 18]. In practice, “likely” identifies patients who should undergo surveillance while confirmation is sought, and current ancillary tests resolve ambiguity in most cases [18].

The author manuscript of the 2022 consensus, deposited in open access in the HAL archive, shows the rationale for the criteria [20]. The working group (13 experts from the IRRDC, C4CMMRD and other centres in Canada, the USA, Austria, Israel and France) drafted the criteria at a workshop in Toronto in October 2017, submitted them to a vote by 60 experts from 18 countries and validated them through surveys with an 80% agreement threshold and through hypothetical cases and real IRRDC families; in that version there were six criteria (four definitive and two probable), compared with seven in the published version reproduced by the AACR, and the group stated that it had not defined criteria to exclude CMMRD, leaving such cases to clinical judgement [15, 20]. The need for criteria combining genetics, ancillary tests and clinical features arises because germline results are inconclusive in a fraction of patients: of 12 families in the Bakry series, 8 had biallelic pathogenic or likely pathogenic variants, 2 had two variants of uncertain significance (VUS), 1 a single pathogenic variant and 1 none; of 38 C4CMMRD families, 28 (74%) had biallelic pathogenic variants and 7 (18%) two VUS; the authors estimate that molecular testing fails to identify biallelic pathogenic variants in up to 40% of patients, and note that PMS2, the gene most often involved, accounts for 49% VUS among reported variants [20]. As for ancillary tests, IHC on non-neoplastic tissue has sensitivity and specificity above 90% in experienced laboratories, but is subjective, requires the pathologist to be aware of the suspicion (the positive internal control is missing) and may be falsely normal with missense variants; gMSI uses dinucleotide microsatellites and is insensitive to MSH6 deficiency; evMSI requires concordance of methylating-agent tolerance and MSI and at least 120 days of culture; and tumour MMR deficiency, TMB and mutational signatures cannot confirm CMMRD, because they also occur in Lynch syndrome and in sporadic tumours with double somatic inactivation, and can only support a diagnosis that already meets the criteria [20]. When results are discordant, an additional ancillary test by an orthogonal method in an accredited laboratory is recommended [20].

The “hallmark cancers” were defined from the IRRDC and C4CMMRD databases: extending the Wimmer tumours (grade III/IV glioma, T-cell non-Hodgkin lymphoma, gastrointestinal adenocarcinoma and polyposis) to all brain and haematological tumours added 18 of 96 IRRDC cases (18.75%) and 31 of 90 C4CMMRD cases (34%) as first diagnosis, and gastrointestinal cancers were the first tumour in 15% and 18%; with the final definition, hallmark cancers account for 76–96% of first diagnoses [20]. The first cancer occurred at age 18 or older in 11–15.6% of cases and at age 25 or older in 1–7%; the latest described was a rectal cancer at age 40 with two PMS2 variants in trans and PMS2 loss in tumour and normal tissue [20]. The consensus also set three practical rules: full siblings of a proband with definitive or probable criteria remain under CMMRD surveillance until the disease is excluded because they have not inherited one of the familial variants (criteria 1, 2 or 5), because they reach age 30 without colorectal polyps or related cancer, or because they have a negative ancillary test when the proband's was positive; CMMRD surveillance is recommended in biallelic deletions of the 3′ region of EPCAM and in the EPCAM/MSH2 combination in trans, until more data are available; and CMMRD surveillance is not recommended in carriers of biallelic MSH3 variants, described with adult adenomatous polyposis, owing to lack of evidence on risk [20].

5.6. Practical algorithm

The algorithm emerging from both guidelines can be summarised in steps: (1) identify the indication (revised C4CMMRD criteria, high TMB below age 18, IHC loss in non-neoplastic tissue, heterozygous MMR variant in a child with cancer, or suspected NF1 meeting selection criteria); (2) assess family history and examine the patient (recommendation 5) [16]; (3) if tumour is available, perform four-protein MMR IHC in tumour and non-tumour cells and, when available, tumour TMB, signature and MSI analysis using sensitive methods [15, 16]; (4) germline testing of all four MMR genes using dedicated PMS2 methods and deletion/duplication analysis, ideally in a panel including POLE, POLD1, EPCAM, APC, MUTYH, NF1 and SPRED1 according to phenotype [15, 17]; (5) parental testing to confirm phase; (6) if inconclusive, validated ancillary testing and transcript analysis [16]; (7) if CMMRD is excluded despite strong suspicion, germline POLE/POLD1 testing [16]. ClinGen classifies mismatch repair cancer syndrome (all four genes, recessive inheritance) as “Strong” actionability in the paediatric setting, justified by the demonstrated survival benefit of surveillance, and “Moderate” in the adult setting [12, 24].

6. Differential Diagnosis

CMMRD differential diagnosis involves three fronts: syndromes producing café-au-lait macules (particularly NF1), other replication repair deficiency syndromes generating hypermutated tumours in children (paediatric Lynch syndrome, polymerase proofreading deficiency, digenic forms and biallelic EPCAM deletion), and childhood adenomatous and hamartomatous polyposis syndromes, which include the historical designation Turcot syndrome [15, 16, 84].

6.1. Neurofibromatosis type 1 and Legius syndrome

Overlap with NF1 is the main cause of diagnostic delay: café-au-lait macules, freckling, neurofibromas and Lisch nodules may occur in CMMRD, and the Oxford Desk Reference NF1 chapter describes it as an “NF1-like” condition with a highly malignant phenotype and autosomal recessive inheritance [1, 74]. A 2017 review showed that any NF1 diagnostic criterion may be present in a patient with CMMRD and that several reported associations between NF1 and high-grade glioma, acute myeloid leukaemia or rhabdomyosarcoma rely on studies that neither proved a germline NF1 variant nor excluded CMMRD and should be reassessed [85]. Features favouring CMMRD are atypical macules (irregular borders, heterogeneous pigmentation), hypopigmented areas, pilomatrixomas, multiple DVAs, consanguinity, familial Lynch syndrome, absence of parental NF1 signs and an atypical NF1 malignancy such as diffuse high-grade glioma or T-lymphoblastic lymphoma; cognitive and learning difficulties are less frequent than in NF1 [15, 16]. Coexistence is possible: patients with CMMRD and NF1 mosaicism exist, so molecular mosaic NF1 diagnosis does not exclude CMMRD with an atypical malignancy [16, 48]. Among children with suspected sporadic NF1 without an NF1 or SPRED1 variant, CMMRD was diagnosed in 0.41%, justifying testing preselected children rather than all children [39, 56]. On brain imaging, multiple DVAs were seen in 83% of children with CMMRD and high-grade glioma and 3% of patients with NF1, and high-grade gliomas were hemispheric in 91% of children with CMMRD versus 25% of the four patients with NF1 and high-grade glioma [16, 62].

6.2. Paediatric Lynch syndrome

MMR-deficient high-grade gliomas in children and young adults more often result from Lynch syndrome than CMMRD: in a multicentre study of 1,389 gliomas in patients aged 0–40 (Toronto, St. Jude and Children's Brain Tumor Network), primary MMR deficiency occurred in 3.7% to 12.4% of high-grade gliomas (30 of 483; 6.2%) and fewer than 1% of low-grade gliomas, with germline aetiology in 33 of 35 cases (94.3%), including unrecognised Lynch syndrome [41]; the 2026 review indicates that of 33 MMR-deficient tumours in this study, 20 were associated with Lynch syndrome and 13 with CMMRD, and that in the German HIT-HGG-2013 trial only one of ten cases was due to CMMRD and six to Lynch syndrome [18]. Distinction relies on IHC (loss only in tumour in Lynch syndrome, also in normal tissue in CMMRD), germline testing with phase analysis and ancillary tests, which do not detect constitutional instability in heterozygotes [15, 80, 82]. Once repair-deficient biology is established in a glioma from a patient with Lynch syndrome, the AACR recommends applying CMMRD therapeutic principles, including maximal resection and immunotherapy, but does not extend intensive childhood surveillance to heterozygotes [15].

6.3. Polymerase proofreading deficiency and digenic forms

Heterozygous variants in POLE and POLD1 exonuclease domains cause polymerase proofreading-associated polyposis, with colorectal and endometrial cancer in adults, but certain variants known as somatic in ultrahypermutated tumours produce a CMMRD-like paediatric phenotype when germline. A 14-year-old boy with polyposis and rectosigmoid carcinoma had POLE variant p.(Val411Leu), multiple café-au-lait macules and a pilomatrixoma [86]; in three other patients with a highly suggestive phenotype, CMMRD was excluded and a heterozygous germline POLE variant identified, and the authors propose CMMRD surveillance protocols for these patients [78]. Another paediatric case with a germline POLE variant, CMMRD-like skin changes and high-grade gliomas was published in 2026 [87]. Digenic combinations of a heterozygous germline POLE or POLD1 variant with a heterozygous MMR variant have also been described [15, 16]. The AACR considers evidence insufficient to offer CMMRD-type surveillance to POLE/POLD1 carriers generally, recommending polyposis guidelines and close monitoring for brain tumours, and proposes individualised plans discussed with the consortia for digenic forms [15].

6.4. Polyposis syndromes and the historical designation Turcot syndrome

The association of colorectal adenomatous polyposis and primary brain tumour, termed Turcot syndrome, is genetically heterogeneous: some cases represent APC-related familial adenomatous polyposis, predominantly with medulloblastoma, and others repair deficiency, predominantly with glioblastoma; the Oxford Desk Reference also notes that the MUTYH contribution remains undetermined and that colorectal polyposis and brain tumours in the first decade are associated with biallelic MMR variants at the same locus [4, 29, 30]. The entities were clinically separated before molecular genetics. Itoh and Ohsato (1985) distinguished, among published cases, a main group with colonic-lesion characteristics of the original Turcot cases (20 to 100 polyps, large polyps over 3 cm in diameter and colon cancer in the second or third decade), a group with too few polyps for polyposis diagnosis and a group with numerous polyps resembling familial adenomatous polyposis [88]. Paraf and colleagues (1997), in a statistical analysis of “brain tumour–polyposis syndrome” (BTP) cases, separated BTP type 1, with malignant glioma before age 20 and colorectal adenomas without polyposis, consanguinity, skin lesions, café-au-lait macules and replication-error tumours, from BTP type 2, with central nervous system tumour (mainly medulloblastoma) in families with familial adenomatous polyposis and germline APC variants; the entities differed in polyp phenotype, colorectal cancer, brain tumour type, skin lesions, macules and consanguinity [89]. BTP type 1 overlaps with current CMMRD, as indicated by the BTPS1 synonym assigned by OMIM to MMRCS1, although some repair-deficient gliomas in young people represent Lynch syndrome (section 6.2); according to OMIM, in recessive forms polyps arise earlier, are fewer and larger, and undergo earlier malignant transformation than in familial adenomatous polyposis [5]. A 1998 review of glioblastoma cases unassociated with familial adenomatous polyposis recorded onset before age 26 and mean survival of 27 months, exceeding sporadic glioblastoma [5]. In children with adenomatous polyps, consider familial adenomatous polyposis (sometimes before age 10), MUTYH-associated polyposis (whose carcinomas may acquire MSI through somatic MMR gene events and mimic Lynch syndrome), NTHL1-associated polyposis and recessive MSH3- or MLH3-related forms [15, 30, 72, 84]. In children, 90% of polyps up to age 10 are juvenile, and juvenile polyposis and Peutz–Jeghers syndromes enter the differential; atypical juvenile polyps have been described in a polyposis-like CMMRD phenotype [66, 84].

6.5. Other differential diagnoses

Other paediatric predisposition syndromes with gliomas, leukaemias or sarcomas, such as Li–Fraumeni syndrome, are distinguished by normal functional-assay repair activity and absence of constitutional instability [44, 80]. In the Dutch T-lymphoblastic lymphoma cohort, the other syndrome identified was Fanconi anaemia (one patient in 88) [40]. Tumours with exclusively somatic MMR deficiency (2.3% of paediatric high-grade gliomas in HIT-HGG-2013, including MSH2 silencing through hypermethylation) and gliomas recurring after temozolomide, acquiring hypermutation with secondary MMR deficiency and COSMIC signature 11, also display hypermutation without germline predisposition [16, 21, 76]. Biallelic EPCAM deletion produces tissue-specific deficiency limited to gastrointestinal tumours [14]. Hypomorphic CMMRD forms with first cancer in adulthood mimic Lynch syndrome and are distinguished only by biallelic genotype and constitutional MSI [54].

7. Treatments and Risk Management Strategies

7.1. General principles

There is no treatment targeting the genetic defect. Management rests on four pillars: surveillance for presymptomatic detection, resection of detected lesions, cancer treatment adapted to repair-deficiency biology and psychosocial and family support [15, 16]. European guidelines recall that several patients with CMMRD have been cured of cancer and that a curative approach should therefore be considered when cancer is diagnosed (strong recommendation); decisions should be discussed at a multidisciplinary meeting involving the treating physician, a specialist in the cancer type and a CMMRD specialist; and patients should be enrolled in clinical trials whenever possible [16]. CMMRD is probably not a contraindication to radiotherapy or haematopoietic stem cell transplantation when indicated (moderate strength) [16]. The AACR adds that, unlike other DNA repair syndromes, MMR deficiency primarily affects replication, preserves normal-tissue response to genotoxic agents and radiotherapy and does not justify reducing standard doses; reirradiation has been described as safe and effective in repair-deficient gliomas, with possible synergy with immunotherapy [15]. The 2026 European review notes that allogeneic transplantation may complicate immunotherapy for subsequent cancers, requiring expert discussion before deciding; the pembrolizumab label warns of serious, including fatal, complications in patients receiving allogeneic transplantation before or after an anti-PD-1/PD-L1 antibody [18, 22].

The frequency of multiple tumours shapes all decisions. Patients may present with several tumours simultaneously or develop them during treatment, so surveillance should continue at diagnosis and during therapy; suspected recurrence should raise the possibility of a second primary tumour, resolved through comparative molecular analysis of original and current samples (identical mutations indicate recurrence; different mutations, a new primary); and fresh tumour samples should be collected and stored with consent [16, 18]. Patients with multiple metachronous cancers may reach maximum cumulative radiotherapy and certain cytotoxic drug doses, reinforcing the value of early detection and treatments that preserve these reserves [15, 37].

7.2. Cancer surveillance

Surveillance begins with shared decision-making: patients and parents should be informed about tumour risks and warning symptoms, discuss advantages and disadvantages of each modality with their clinician and be encouraged to share results in research registries [16]. The table compares European (ERN GENTURIS/C4CMMRD, 2024) and North American (AACR, 2024) recommendations.

Sources: ERN GENTURIS [16]; AACR [15]; comparison and practical notes from the 2026 European review [18].

The rationale for each difference is explicit in the sources. European guidelines abandoned routine abdominal ultrasound and full blood count because, in both prospective cohorts, haematological malignancies were detected mainly incidentally or between examinations, because T-lymphoblastic lymphoma is more often mediastinal and invisible on abdominal ultrasound, and because rapid growth would require 3-month intervals; repeated chest radiographs are not recommended given potential genotoxicity [16]. The AACR retained six-monthly abdominal ultrasound and left full blood count and lactate dehydrogenase decisions to physician and family given insufficient robust data [15]. Circulating T-cell rearrangement testing remains under investigation [16]. For adults, no brain tumour risk studies exist because most patients did not reach adulthood before surveillance programmes; annual intervals after age 20 are a consensus recommendation [16]. European guidelines do not recommend endometrial biopsy because of pain and added burden, or urinary cytology, whose benefit has not been demonstrated in Lynch syndrome [16].

The first published systematic surveillance experience came from the international consortium in 2014: Bakry and colleagues used the protocol described by Durno and colleagues in 2012, including six-monthly brain MRI, annual gastrointestinal endoscopy and full blood count, erythrocyte sedimentation rate and lactate dehydrogenase testing every 4 months; 39 asymptomatic lesions were detected, including 2 malignant gliomas, 2 gastrointestinal carcinomas and 6 gastrointestinal polyposis cases, all tumours were completely resectable and, after mean follow-up of 61 months, all protocol patients were alive [5, 51]. Without a comparison group and with few patients, this historical result documented feasibility and protocol yield without estimating survival benefit; periodic blood tests in that protocol were later abandoned by the 2024 European guidelines and retained only as an AACR option (table above) [15, 16]. Surveillance efficacy was demonstrated in subsequent observational studies. In the IRRDC cohort, 89 patients followed prospectively from 2008 were classified as full (33), partial (20) or no surveillance (36): 4-year overall survival was 79% (95% CI 54.8–90.9), 55% (28.5–74.5) and 15% (5.2–28.8), respectively (P < 0.0001), and surveillance was the only variable associated with better survival in multivariable analysis; per tumour, 5-year overall survival was 90% for asymptomatically detected cancers and 50% for symptomatic cancers [37]. In patients under surveillance, all gastrointestinal tumours (24), all other solid tumours (5) and 75% of brain tumours (15 of 20) were detected asymptomatically, but only 2 of 12 haematological malignancies (16%), and the five symptomatic brain tumours occurred with intervals exceeding 6 months or surveillance interruptions; 43% of deaths under surveillance were due to haematological malignancies [37]. To mitigate lead-time bias, the authors compared mortality at median follow-up of 25.4 months (30% with full surveillance versus 72% without; P = 0.001) [37]. In the European registry, 22 definitively diagnosed patients were followed for a mean of 4 years: the programme detected 8 asymptomatic malignant tumours (3 brain, 3 upper gastrointestinal, 2 colorectal) and numerous adenomas, 7 patients developed symptomatic malignancies (2 brain, 1 small bowel and 4 haematological) and 16 (73%) were alive at follow-up end [90]. Partial surveillance was also associated with benefit, relevant to low- and middle-income countries [18, 37]. These observational studies, with adherence decided by patients and clinicians, cannot precisely estimate each modality's isolated benefit; whole-body MRI detected only one malignant tumour in the 2021 cohort [37].

7.3. Tumour treatment

High-grade gliomas. Surgery remains the main treatment for resectable gliomas [16]. Temozolomide should be avoided (strong recommendation), because MMR deficiency is a major resistance mechanism and glioblastomas recurring after temozolomide frequently acquire hypermutation with MMR deficiency [16]. Immunotherapy with a PD-1 inhibitor should be considered, preferably in a clinical trial (strong) [16]. Conventional treatment outcomes are poor: in the European registry, 3-year overall survival in high-grade gliomas was 20.5%, and eight patients treated with immunotherapy for recurrence progressed within the first 2 months, with six deaths at a median of 5.2 months [18, 61]. In the multicohort study of primary MMR-deficient gliomas in patients up to age 40, checkpoint blockade was associated with better survival than conventional chemoradiotherapy (HR 0.4; 95% CI 0.3–0.7), irrespective of age and germline aetiology [41]. Low-grade gliomas should be resected whenever possible without excessive neurological risk (strong); first-line medulloblastoma treatment should probably not differ from sporadic disease (moderate) [16].

Lymphomas and leukaemias. Chemotherapy should probably resemble that used without CMMRD; in a second primary lymphoma, first-line treatment adapted to subtype, accounting for cumulative doses already received, should be preferred over a relapse regimen (moderate recommendations) [16]. Despite in vitro thiopurine tolerance of MMR-deficient cells, the international series of 74 patients showed no increased failure of standard regimens: after first lymphoma, 5-year event-free survival was 23.5% and overall survival 61.5%, and 5-year cumulative risk of progression or relapse, second malignancy or death as first event was 20.8%, 52.9% and 2.7%, meaning prognosis is limited mainly by subsequent malignancies [16, 65]. In the Dutch cohort, 6 of 8 patients with CMMRD achieved complete remission after T-lymphoblastic lymphoma treatment according to EURO-LB02 [40]. In an earlier French series, failure appeared higher than expected in T-cell non-Hodgkin lymphomas (progression or relapse in 6 of 12) [91]. Checkpoint inhibitors produced no responses in haematological malignancies treated in the IRRDC registry (0 of 3), and the AACR describes lack of efficacy in T-cell malignancies [15, 21].

Lynch-spectrum tumours and polyposis. Lynch-spectrum cancers in patients with CMMRD should be treated according to guidelines for Lynch-associated tumours (strong); immunotherapy should be first-line treatment for bulky, unresectable or metastatic colorectal tumours (strong) and, ideally within trials, extracolorectal Lynch-spectrum tumours (strong) [16]. For tumours outside the Lynch spectrum, immunotherapy should be encouraged through interdisciplinary discussion at any stage, particularly when standard options offer a low probability of cure (moderate) [16]. Patients with multiple colorectal adenomas should receive surgery according to general practice for other polyposis syndromes (strong), but European guidelines do not recommend systematic early gastrointestinal surgery: polyp numbers are usually endoscopically manageable, immunotherapy is effective in gastrointestinal carcinomas, extensive surgery reduces quality of life and prognosis depends mainly on brain tumours; decisions should be based on endoscopic control, with six-monthly endoscopy if needed [16]. In adults, approximately 5% of metastatic colorectal carcinomas are dMMR/MSI-H and eligible for first-line checkpoint inhibitors (pembrolizumab or nivolumab with ipilimumab), supporting extrapolation of this approach [92]. The AACR considers total colectomy with pouch worth discussing for numerous individually unresectable polyps, on the same grounds [15]. In the 2026 report of a 10-year-old child with hepatic flexure adenocarcinoma, duodenal adenoma and biallelic PMS2 variants, neoadjuvant ipilimumab and nivolumab were used before total abdominal colectomy and adjuvant CAPEOX, with partial pathological response at stage IIIB [93].

Immunoglobulins. Patients with reduced IgG/IgA levels or deficiency should not receive replacement treatment in the absence of clinical manifestations (strong) [16].

7.4. Immune checkpoint inhibitors

The biological basis is hypermutation, generating neoantigens and a lymphocyte-infiltrated microenvironment even in usually “cold” tumours such as gliomas [21, 46]. The first report, in 2016, described two siblings with recurrent multifocal glioblastoma treated with nivolumab, with clinical and radiological responses; the same study showed that biallelic-deficient glioblastomas had neoantigen loads 7 to 16 times those of melanomas, lung cancers or MSI gastrointestinal cancers responsive to immunotherapy [38]. The main clinical data are summarised in the table.

These findings require caution. Most derive from observational registries without randomised control groups, with treated-patient selection, drug choice and duration determined by clinicians, and overlapping patients across publications from the same consortium [21, 96]. The only prospective trial included just 10 evaluable patients and was terminated prematurely [94, 95]. The discrepancy between European and North American recurrent-glioma results may reflect selection, era and biological differences: in 2026, gliomas with secondary polymerase mutations were shown to be ultrahypermutated and responsive, whereas IDH1-mutant gliomas were immunologically “cold” and resistant [47]. High TMB and MSI were independent monotherapy response biomarkers [15, 21]. Delayed responses are common: pseudoprogression (flare) occurred at a median of 34 days with acute clinical deterioration; eight patients stopping treatment died, and four continuing with appropriate support achieved responses [21]. Comparing response rates across these studies is inappropriate because populations, inclusion criteria and response definitions differ.

Doses and regulatory status. No checkpoint inhibitor has an approved indication specifically for CMMRD. In the United States, pembrolizumab (Keytruda®) received accelerated approval on 23 May 2017 for adult and paediatric patients with unresectable or metastatic MSI-H or dMMR solid tumours progressing after prior treatment and without satisfactory alternatives, based on an overall response rate of 39.6% (95% CI 31.7–47.9) in 149 patients with 15 tumour types; the approval letter included the limitation that safety and efficacy in children with MSI-H central nervous system cancers were not established [97, 98]. In the consulted label version (July 2026), the indication includes adults and children with MSI-H/dMMR solid tumours determined by an FDA-authorised test, with a paediatric dose of 2 mg/kg (up to 200 mg) every 3 weeks, and the paediatric central nervous system limitation of use refers to TMB-high cancers [22]. Nivolumab (Opdivo®) was approved in 2017 as monotherapy and in 2018 combined with ipilimumab (Yervoy®) for MSI-H/dMMR metastatic colorectal cancer in adults and children aged 12 or older [99–101]. In the European Union, MSI-H/dMMR indications for pembrolizumab (colorectal; endometrial, gastric, small bowel and biliary tract) and nivolumab with ipilimumab (colorectal) are exclusively for adults, and the pembrolizumab SmPC states that safety and efficacy below age 18 are not established except in melanoma and classical Hodgkin lymphoma [23, 102]. Dostarlimab (Jemperli®) has a US adult dMMR solid tumour indication and an EU endometrial indication, without established paediatric data [103, 104]. Use in paediatric gliomas, children younger than 12 (nivolumab) or any paediatric indication in the EU is therefore off-label, supported by the cited guidelines and studies.

Sources: FDA labels consulted through openFDA and Drugs@FDA [22, 98–101, 103]; EMA data and SmPCs [23, 102, 104]. Authorisation does not imply reimbursement or access in each country; the sources consulted do not include INFARMED or ANVISA/CONITEC information on this use.

IRRDC glioma salvage doses were nivolumab 3 mg/kg every 2 weeks (maximum 240 mg per dose) or pembrolizumab 2 mg/kg every 3 weeks (maximum 200 mg), combined with ipilimumab 1 mg/kg every 3 weeks for 4 doses or, in RAS-MAPK-altered tumours, trametinib 0.025 mg/kg daily (maximum 2 mg per dose); these combinations are off-label and have not been evaluated in randomised trials [96]. These are the reference doses; the labels serve for comparison. Nivolumab 3 mg/kg every 2 weeks and pembrolizumab 2 mg/kg every 3 weeks (maximum 200 mg) match the paediatric doses in the FDA labels [22, 101]. Ipilimumab 1 mg/kg every 3 weeks, for a maximum of 4 doses, with nivolumab 3 mg/kg (below 40 kg) or 240 mg (40 kg or more), is the FDA label dose in MSI-H/dMMR colorectal carcinoma from age 12; in the EU, this combination in colorectal carcinoma is authorised only in adults [105, 106]. Trametinib (Mekinist®) has a paediatric indication only in combination with dabrafenib in tumours with BRAF p.(Val600Glu), with dosing by weight band: 1 mg tablets (26–37 kg), 1.5 mg (38–50 kg) or 2 mg (51 kg or more) once daily, and, in the USA, oral solution from 8 kg (0.3 mg at 8 kg, approximately 0.033–0.038 mg/kg); in the European SmPC, the tablet dose below 26 kg is not established [107, 108]. The dose recommended by the consortium, 0.025 mg/kg/day (maximum 2 mg), is lower than the label doses, which refer to another indication, and is, for information, off-label [96]. Regarding duration, in the report of two siblings nivolumab was continued for 24 months and interruption correlated clinically with recurrence on several occasions, without an established optimal duration [60].

Adverse effects and management. PD-1 inhibitors may cause severe or fatal immune-mediated reactions in any organ, including pneumonitis, colitis, hepatitis, endocrinopathies, nephritis and skin reactions; the label recommends assessing liver enzymes, creatinine and thyroid function at baseline and periodically, withholding or permanently discontinuing according to severity and, when required, administering systemic corticosteroids (prednisone 1 to 2 mg/kg/day or equivalent, with no maximum dose defined in the source) until improvement to grade 1 or less [22]. In 173 children in KEYNOTE-051 treated with 2 mg/kg (maximum 200 mg) every 3 weeks, adverse reactions occurring at least 10% more frequently than in adults were pyrexia (33%), vomiting (29%), headache (25%) and abdominal pain (23%), alongside haematological changes [22]. In CMMRD, normal cells accumulate microsatellite mutations and indels, and high autoimmune toxicity rates have been described with combinations: refractory myositis after adding trametinib, cardiac toxicity with trametinib, autoimmune pneumonitis with nivolumab and ipilimumab and severe hepatotoxicity with ipilimumab [60, 96]. Pseudoprogression should be distinguished from progression before stopping treatment [21].

7.5. Prevention

Acetylsalicylic acid (ASA) chemoprevention reduced colorectal carcinoma incidence in Lynch syndrome carriers in the CAPP2 trial and might benefit patients with CMMRD, but strong evidence of reduced cancer incidence in CMMRD is lacking; European guidelines recommend discussing benefits and potential adverse effects with patients (recommendation 22, moderate strength) and state that evidence is insufficient to recommend its use as chemoprevention [16, 109]. The 2026 European review interprets this position as a non-recommendation and identifies higher brain tumour risk and cerebral DVAs as reasons for controversy; there is no label for cancer chemoprevention [18]. The reference dose is that of the CAPP2 trial, 600 mg/day (fixed daily dose in adults; no maximum dose defined in the source) for a median of 25 months in adults with Lynch syndrome, and no dose has been studied in children with CMMRD [17, 109]. For comparison, the Portuguese SmPC for low-dose ASA (Aspirina GR® 100 mg) indicates 100–300 mg/day only for antiplatelet therapy in adults, so that the CAPP2 dose is off-label; the SmPC states that it should not be used below 18 years of age and warns against the use of ASA-containing medicines in children and adolescents with viral infections, owing to the risk of Reye syndrome [110]. Frameshift neoantigen vaccines demonstrated safety and immunogenicity in a phase I/IIa trial in patients with MMR-deficient colorectal cancer (22 patients) and are proposed as a possible future CMMRD prevention strategy; the consortia are also developing an anti-PD-1 trial to delay or prevent a second malignancy [16, 18, 111].

7.6. Ongoing trials and developments in the last 12 months

Sources: ClinicalTrials.gov [95, 112–116]. The IRRDC trial page still described the combination trial as open, which the official registry no longer confirms [27].

Between October 2025 and October 2026, developments identified were scientific and registry-related rather than regulatory: classification of primary MMR-deficient gliomas into three subgroups with different immunotherapy responses (peer-reviewed article) [47]; characterisation of primary MMR-deficient IDH-wildtype and H3-wildtype gliomas as a distinct type with potential prolonged survival [63]; the European review on international networks' impact, with a comparative surveillance table [18]; registration of the phase II relatlimab and nivolumab trial, still without patients [113]; the comprehensive Lynch syndrome GeneReviews update of 17 September 2026 retaining CMMRD as an allelic disorder [17]; and neoadjuvant immunotherapy case reports in paediatric colorectal carcinoma [93]. No CMMRD-specific authorisation applications, CHMP opinions or FDA decisions were found in the sources consulted; this negative search does not prove that unpublished processes do not exist.

8. Reproductive Options and Family Counselling

8.1. Recurrence risks and implications for relatives

CMMRD is inherited in an autosomal recessive manner. When both parents are heterozygous in the same gene, each pregnancy has a 25% probability of CMMRD, 50% of Lynch syndrome (heterozygosity, like the parents) and 25% of inheriting neither variant [17, 27]. The family situation is therefore dual: the child has CMMRD and requires intensive paediatric surveillance, while parents and many relatives on both sides carry Lynch syndrome, with gene-specific adult risks and surveillance [16, 27]. The Oxford Desk Reference emphasises that CMMRD families, frequently consanguineous, are now well described and that the phenotype differs from Lynch syndrome [4].

European guidelines make twelve counselling recommendations [16]:

The quantitative rationale for partner testing was given for PMS2: with population carrier frequency of 1 in 714, the a priori probability of a PMS2 Lynch syndrome carrier having a child with CMMRD is approximately 1 in 2,856 (1/4 × 1/714), and that for a patient with PMS2-related CMMRD is 1 in 1,428 (1/2 × 1/714), with higher values in consanguinity or founder populations [16]. This topic did not reach consensus at a C4CMMRD meeting in 2022 because it depends on testing access in each country, because complete gene analysis in a partner may reveal variants of uncertain significance complicating counselling, and because complete PMS2 analysis remains complex and sparsely available; recommendations 9 to 11 resulted from the Delphi process [16]. GeneReviews records the same situation from the Lynch syndrome perspective: if a carrier's reproductive partner has a variant in the same gene, children may inherit biallelic variants and have CMMRD [17]. A 2025 report described a couple using in vitro fertilisation unaware that both carried Lynch syndrome, whose child, healthy at birth, developed acute bilineal leukaemia and was diagnosed with CMMRD; the authors advocate including Lynch syndrome genes in prenatal genetic testing, a position not corresponding to guideline recommendations for couples without risk factors [16, 117].

The AACR adds practical points: all families should receive genetic counselling; when proband testing is non-diagnostic, cascade testing should use functional assays in the proband and then relatives; sibling testing in a child with CMMRD may reveal Lynch syndrome; and, in adult-onset syndromes such as Lynch syndrome, testing asymptomatic siblings should generally be deferred until adulthood or until it affects management [15]. A tension therefore exists between systematic sibling CMMRD testing (European recommendation 4) and deferral of knowledge of Lynch carrier status, addressed through case-by-case discussion in European recommendation 5 [15, 16].

8.2. Reproductive options

Options include natural conception accepting the risk, prenatal diagnosis, preimplantation genetic testing for monogenic disorders, gamete donation, adoption and postnatal predictive testing [118]. The 2026 ERN GENTURIS reproductive counselling guideline for cancer predisposition syndromes recommends proactively offering counselling with psychological support and multidisciplinary collaboration, explaining each option's benefits, limitations, success rates and waiting times and respecting national legislation; invasive prenatal diagnosis (estimated miscarriage risk below 0.5%) is not recommended without intention to terminate pregnancy, if the couple cannot bear the emotional burden of that decision or if the law prohibits it; and, when predisposition manifests in childhood or requires childhood surveillance, predictive testing can occur at the age surveillance is indicated [118]. In CMMRD, where brain surveillance may begin at birth or by age 2, this means early testing; the AACR adds that, if testing is not undertaken before or during pregnancy, it should occur after birth, including through functional assays, to avoid unnecessary surveillance of unaffected children [15, 16]. Prenatal and preimplantation options are limited or complicated when proband germline testing has not identified the variants [15].

For women with CMMRD, European guidelines recommend annual gynaecological surveillance from age 20 and discussing prophylactic hysterectomy after completing childbearing [16]. The sources consulted contain no specific pregnancy recommendations for women with CMMRD; PD-1 inhibitor labels warn of embryo-fetal toxicity and recommend effective contraception in women of reproductive potential during treatment [22, 101].

8.3. Psychosocial support

The four European quality-of-life recommendations are all strong: offer psychological support to patients and families during diagnostic assessment and at any point during treatment and surveillance; offer age-appropriate education; and ensure professionals address the diagnosis's psychosocial implications [16]. Guidelines acknowledge that awareness of high risk can increase distress before and after investigations and that detection of small lesions of uncertain significance requiring short-term surveillance increases anxiety [16].

9. Special Considerations: Population Diversity and Epidemiology

9.1. Frequency

CMMRD birth incidence is estimated at approximately 1 in 1,000,000 when parents are unrelated, supported by empirical screening data in children with suspected NF1 [16, 56]. Orphanet classifies prevalence as below 1 in 1,000,000 [3]. Lynch syndrome is estimated to be approximately 3,000 times more frequent than CMMRD [18]. Hospital cohort frequencies are not population prevalences: they measure CMMRD proportions in patient groups already selected for a tumour type or clinical sign, as summarised below.

9.2. Consanguinity, endogamy and founder effect

Approximately half of patients with CMMRD are homozygous, and prevalence is higher in populations with consanguinity, endogamy or founder variants [16]. In the 2014 review, parental consanguinity or homozygosity for a founder variant occurred in 46 of 91 families [1]; consanguinity was recorded in 9 of 23 French families [91], 11 of 14 families in the first international consortium report [51], 20 of 38 families in the European brain tumour registry [61] and 54% of lymphoma patients [65]. In low-consanguinity countries, compound heterozygosity predominates, but homozygosity may be common in minorities, Indigenous communities, closed communities and immigrants, so diagnosis should not depend on consanguinity history [15, 19].

Specific variants and populations have been described. In Israel, Ashkenazi Lynch syndrome founder variants were found in two non-consanguineous Ashkenazi families, homozygous in one and compound heterozygous in the other, and four consanguineous families of various ancestries had different homozygous variants [122]. In Nunavik (Quebec), PMS2 c.2002A>G is an Inuit founder variant associated with attenuated phenotype [53]; according to OMIM, the authors estimated an allele frequency of 1 in 16 individuals in the Nunavik Inuit population and dated founder origin to the late 11th century [11]. In the United Kingdom, 13 patients from six consanguineous families of Pakistani origin had PMS2-related CMMRD, and five of these families shared NM_000535.7:c.2404C>T p.(Arg802Ter) (historical designation R802X), with founder-effect evidence [57, 58]; the sixth family had, according to OMIM, homozygosity for a 1 bp deletion designated Y181X (p.Tyr181Ter), whose complementary DNA notation is absent from the source consulted [11]. Other OMIM-summarised reports involve Pakistani families with MSH2 and MSH6, Turkish families with MLH1 and PMS2, a Guyanese family with PMS2 and a patient of Moroccan ancestry with MSH6 loss in tumour and normal tissue (several in section 3.9) [5, 9–11]. In Quebec's French Canadian population, PMS2 c.2117del was identified in 22 families (21 with Lynch syndrome and 1 with CMMRD), at a frequency of 0.35% in 6,347 newborns and 0.12% in 18,129 CARTaGENE cohort participants, and MSH6 c.10C>G affects 1 in 400 French Canadians [56, 59]. PMS2 c.736_741delinsTGTGTGTGAAG is a founder variant prevalent in Europe and North America [56]. No PMS2 or MSH6 founder variants are described in the Netherlands, but two recurrent variants (PMS2 c.137G>T and MSH6 c.651dup) occurred in the lymphoma cohort [40]. In Jordan, where consanguinity is high, MMR deficiency was found in 39% of high-grade gliomas and 44% of paediatric colorectal carcinomas [119, 123].

9.3. Europe, United States, Portugal and Brazil

The two main data sources are North American and European: the Toronto-based IRRDC cohort with patients from more than 50 countries [19], and the Paris C4CMMRD registry with more than 100 patients, mainly European [18]. The French series of 31 patients diagnosed in French laboratories had variants in PMS2 (18), MSH6 (6), MLH1 (4) and MSH2 (3), 67 tumours and 22 deaths, 9 (41%) from the primary tumour [91]. IRRDC lists Portugal and Brazil among countries with collaborating centres [27]. The sources consulted contain no incidence data or CMMRD series from Portugal or Brazil; GeneReviews lists Lynch syndrome founder variants described in Portugal (deletion of MLH1 exons 17–19 with LRRFIP2 deletion; MSH2 c.2152C>T), Portugal and South America (MSH2 c.388_389del), and MLH1 c.545+3A>G in Italy, Canada and Brazil [17]. Their CMMRD relevance follows the principle that founder populations increase the probability of couples carrying variants in the same gene, but their actual contribution to CMMRD cases in these countries has not been quantified in the sources consulted [16].

9.4. Resource-limited settings

CMMRD is concentrated in high-consanguinity countries, many with limited resources, where access to complex germline testing, frequent magnetic resonance imaging and checkpoint inhibitors is restricted; in Pakistan, the authors emphasise the need for protocols adapted to low- and middle-income countries given immunotherapy costs [120]. Genomic MSI functional assays costing less than 150 dollars per sample, and evidence that partial surveillance also improves survival, offer realistic paths in these settings [15, 18, 37].

10. Illustrative Clinical Case

The case published by Palova and colleagues in 2024 describes two brothers from a non-consanguineous family without a cancer history, compound heterozygous for two PMS2 variants inherited from asymptomatic parents: NM_000535.7:c.2T>A, published as p.(M1K), affecting the initiation codon and predicted by VariantValidator as p.(Met1?), and NM_000535.7:c.2521del p.(Trp841GlyfsTer10) [13, 60].

The older brother, without skin macules or other neurocutaneous stigmata, had colon adenocarcinoma at age 14, treated with hemicolectomy and FOLFOX, achieving complete remission. Four years later a paediatric-type diffuse high-grade glioma arose (without histone or IDH mutation, with 13q loss). CMMRD was diagnosed by germline sequencing. Both tumours were hypermutated, with MMR-deficiency signatures, PMS2 loss on IHC and high exome MSI burden (colon 207; glioma 1,360; median repair-proficient controls 17); the glioma had TMB 337.56 mutations/Mb and a pathogenic somatic POLE exonuclease-domain variant (p.E978G), with SBS10 signature [60]. Following focal radiotherapy, he entered a dendritic cell vaccine trial (5 doses) and began nivolumab; temozolomide, initially started, was stopped after multidisciplinary discussion. He achieved complete remission and continued nivolumab for 24 months. Five months after discontinuation, disseminated recurrence arose, again hypermutated and MSI-high (TMB 461.87 mutations/Mb), with the same POLE variant but an evolved mutational spectrum and no known immune escape mechanisms. He received focal radiotherapy and bevacizumab, resumed nivolumab and received ipilimumab after radiotherapy, stopped for thrombocytopenia attributed to parvovirus B19 infection; status epilepticus followed with disseminated progression and death 43 months after glioma diagnosis [60]. Figure 1 of the article shows adenocarcinoma and glioma histology, copy-number profile, both PMS2 variants in the genomic viewer, mutational signatures and the patient's treatment and imaging timeline.

The younger brother, diagnosed by cascade testing, had café-au-lait macules and began the published surveillance protocol. Surveillance imaging detected asymptomatic abdominal lymphadenopathy, representing high-grade B-cell lymphoma with 11q aberration without MYC rearrangement. He received two intensive chemotherapy cycles according to NHL-BFM 2012 with rituximab; given hypermutation (TMB 107.74 mutations/Mb), MMR-deficiency signatures, PMS2 loss and high MSI, chemotherapy was stopped and nivolumab begun by multidisciplinary consensus. After more than 24 months of remission, the dose was reduced to 1 mg/kg every 2 weeks; nine months later surveillance MRI detected a putamen and globus pallidus lesion, confirmed by biopsy as paediatric-type diffuse high-grade glioma, with TMB 142 mutations/Mb and driver variant POLE p.V411L. He received cytoreduction, reirradiation and full-dose nivolumab (3 mg/kg every 2 weeks); initial radiological progression prompted trametinib addition for a tumour truncating NF1 variant, stopped for cardiac toxicity; adding ipilimumab to nivolumab led to complete remission, but immunotherapy was stopped for autoimmune pneumonitis, and an attempt to resume ipilimumab alone was stopped for severe hepatotoxicity. At publication, he had no evidence of cancer 13 months after ending immunotherapy, more than 23 months after glioma diagnosis and 5 years after lymphoma [60]. Figure 2 of the article shows lymphoma and glioma morphology with bizarre multinucleated cells, 11q aberrations, mutational signatures and timeline.

The case illustrates several points in this article. First, intrafamilial variability: the brothers share a genotype but only one had café-au-lait macules [60]. Second, absence of consanguinity and family history, typical of PMS2-related CMMRD [16]. Third, surveillance value: both cancers in the younger brother were detected without symptoms, which the authors relate to his better outcome [60]. Fourth, the need for repeat biopsy and genomic analysis when recurrence is suspected, because the mutational spectrum evolves rapidly [16, 60]. Fifth, immunotherapy use in B-cell lymphoma to limit intensive chemotherapy exposure, contrasting with absent T-cell lymphoma responses [21, 60]. Some management decisions in this report, such as nivolumab dose reduction, dendritic cell vaccination and trametinib and ipilimumab use, are individual or research decisions rather than routine recommendations; European guidelines recommend immunotherapy in high-grade gliomas, preferably in a clinical trial, and standard lymphoma treatment [16].

11. Prognosis and Long-Term Follow-up

11.1. Prognosis

Without intervention, most patients do not survive the first decade [18]. In the 2024 IRRDC cohort, 10-year overall survival after diagnosis was 39% (95% CI 30–52) for central nervous system tumours, 67% (55–82) for haematological tumours, 89% (81–97) for gastrointestinal tumours and 96% (88–100) for other solid tumours; at age 15, overall survival was 63% for PMS2, 49% for MSH6, 19% for MLH1 and 0% for MSH2 [19]. Brain tumours are the main childhood cause of death alongside haematological malignancies [15]; in the European registry, survival after first brain tumour was 22% at 5 years [61]. In lymphomas, 5-year overall survival was 61.5%, limited by second malignancies [65]. Surveillance (4-year overall survival 79% with full surveillance versus 15% without) and immunotherapy (estimated 3-year overall survival 41.4% in refractory tumours) have changed prognosis, and increasing numbers reach adulthood, making the adult tumour spectrum an area still being characterised [16, 21, 37]. These figures derive from diagnosed-patient cohorts with heterogeneous follow-up and treatment eras and should not be read as fixed individual risk; hypomorphic-variant carriers may have very different courses [54].

11.2. Follow-up plan

Follow-up combines the tables in section 7.2 with the following practical rules taken from the guidelines: clinical examination every 6 months throughout life; brain MRI every 6 months until age 20 then at least annually (or six-monthly from diagnosis according to the AACR); annual gastrointestinal endoscopies and video capsule from ages 6 to 10, shortening to 6 months with polyps and never exceeding 12 months; annual gynaecological and abdominopelvic surveillance from age 20 (or from age 12 according to the AACR); whole-body MRI at least once; annual skin examination (AACR); and continued surveillance during and after treatment of any cancer [15, 16]. Any finding should prompt earlier repeat examination according to clinical judgement [15]. Detected benign or low-grade lesions (polyps, low-grade gliomas, pilomatrixomas, bone tumours) should whenever possible be resected before accumulating mutations and transforming [15]. Patients should be followed at experienced centres with access to specialised paediatric endoscopy, and transition to adult care must preserve brain and gastrointestinal surveillance because data on adult risk remain insufficient [16, 18]. Heterozygous parents and relatives should be followed according to the Lynch syndrome guidelines for the gene concerned [16, 17].

11.3. Reference networks, registries and resources

In Europe, CMMRD belongs to ERN GENTURIS thematic group 4 (“other rare, predominantly malignant genetic tumour risk syndromes”), which published the 2024 guidelines and lists CMMRD among its clinical practice guidelines [16, 25]. Paediatric patients with cancer also benefit from ERN PaedCan, the European paediatric oncology network [26]. C4CMMRD maintains a European registry in Paris, and IRRDC, based at Toronto's Hospital for Sick Children, collaborates with more than 100 hospitals in 46 countries, provides patient and family pages and a surveillance protocol, and facilitates diagnostic testing access [18, 27]. The AACR recommends informing patients and families that these consortia provide information and support [15]. The sources consulted did not identify a CMMRD-specific patient organisation; for Lynch syndrome relatives, the Oxford Desk Reference lists Lynch Syndrome UK (part of Lynch Syndrome International) and the International Society for Gastrointestinal Hereditary Tumours (InSiGHT) [4]. European guidelines were developed with patient-representative participation [16].

12. Summary and Final Considerations

CMMRD results from biallelic germline inactivation of MLH1, MSH2, MSH6 or PMS2 and is an autonomous recessive paediatric entity distinct from heterozygous Lynch syndrome. Absent repair in all cells from conception produces hypermutated and frequently ultrahypermutated tumours, particularly high-grade gliomas, T-lymphoblastic lymphomas and Lynch-spectrum carcinomas, with 90% cumulative cancer incidence by age 18 in diagnosed cohorts, multiple tumours and non-neoplastic manifestations mimicking NF1. Approximately two-thirds of cases are due to PMS2, whose low-penetrance heterozygosity explains frequent absence of family history, and half involve homozygosity associated with consanguinity, endogamy or founder variants. “Turcot syndrome” is a historical heterogeneous designation encompassing familial adenomatous polyposis with medulloblastoma (BTP type 2) and repair deficiency with glioblastoma (BTP type 1, OMIM synonym BTPS1).

Diagnosis rests on C4CMMRD criteria revised in 2024, MMR protein IHC in tumour and normal tissue, germline testing with dedicated PMS2 methods and phase analysis, and ancillary constitutional instability tests achieving sensitivity and specificity close to 100% in published cohorts, except gMSI in MSH6 deficiency. European and North American 2024 guidelines converge on core surveillance (six-monthly clinical examination and brain MRI, annual endoscopies from age 6) and differ on abdominal ultrasound, full blood count, whole-body MRI and starting age for genitourinary surveillance. Surveillance is associated with better survival for brain and gastrointestinal tumours but does not detect haematological malignancies early. In treatment, temozolomide should be avoided in gliomas, radiotherapy and standard chemotherapy are not contraindicated, and PD-1 inhibitors show durable responses in gliomas and solid tumours but not T-cell lymphomas; no drug has an approved CMMRD indication, and paediatric use is largely off-label, particularly in the European Union. Counselling involves the entire family: parents and many relatives have Lynch syndrome, siblings should be tested, and partner testing for Lynch carriers is recommended with consanguinity, founder effect or suggestive family history. Gaps persist: unbiased risk estimates, adult risk, whole-body MRI efficacy, early haematological malignancy detection, prevention and equitable access in lower-resource countries where disease is more frequent.

13. Bibliographic References

1. Wimmer K, Kratz CP, Vasen HF, Caron O, Colas C, Entz-Werle N, et al. Diagnostic criteria for constitutional mismatch repair deficiency syndrome: suggestions of the European consortium 'care for CMMRD' (C4CMMRD). J Med Genet. 2014;51(6):355-65. doi:10.1136/jmedgenet-2014-102284. PMID:24737826.
2. Durno C, Boland CR, Cohen S, Dominitz JA, Giardiello FM, Johnson DA, et al. Recommendations on Surveillance and Management of Biallelic Mismatch Repair Deficiency (BMMRD) Syndrome: A Consensus Statement by the US Multi-Society Task Force on Colorectal Cancer. Gastroenterology. 2017;152(6):1605-1614. doi:10.1053/j.gastro.2017.02.011. PMID:28363489.
3. Orphanet. Constitutional mismatch repair deficiency syndrome. ORPHA:252202 [Internet]. Disease information. Prevalence <1/1 000,000; ICD-10 D80.8; ICD-11 LD2D.10. Version 1.6.9; accessed 7 Oct 2026.
4. Firth HV, Hurst JA. Oxford Desk Reference: Clinical Genetics and Genomics. 2nd ed. Oxford: Oxford University Press; 2017. Chapter 4, Cancer: Lynch syndrome (LS). p. 586–591.
5. Online Mendelian Inheritance in Man, OMIM®. #276300 Mismatch repair cancer syndrome 1; MMRCS1 [Internet]. Baltimore (MD): Johns Hopkins University. Created: Victor A. McKusick, 4 Jun 1986; last updated: Kelly A. Przylepa, 18 May 2021 (edited 19 May 2021). OMIM 276300. Accessed 7 Oct 2026.
6. Orphanet. Non-polyposis Turcot syndrome. ORPHA:99817 [Internet]. Record. Entity removed from nomenclature and transferred to Lynch syndrome. Accessed 7 Oct 2026.
7. Orphanet. Turcot syndrome with polyposis. ORPHA:99818 [Internet]. Record. Entity removed from nomenclature and transferred to familial adenomatous polyposis. Accessed 7 Oct 2026.
8. National Center for Biotechnology Information. MedGen: Mismatch repair cancer syndrome 1 (C5399763; OMIM 276300; MLH1), 2 (C5436806; OMIM 619096; MSH2), 3 (C5436807; OMIM 619097; MSH6) and 4 (C5436817; OMIM 619101; PMS2) [Internet]. MedGen. Accessed 7 Oct 2026.
9. Online Mendelian Inheritance in Man, OMIM®. #619096 Mismatch repair cancer syndrome 2; MMRCS2 [Internet]. Baltimore (MD): Johns Hopkins University. Created: Anne M. Stumpf, 13 Nov 2020 (edited 24 Nov 2020). OMIM 619096. Accessed 7 Oct 2026.
10. Online Mendelian Inheritance in Man, OMIM®. #619097 Mismatch repair cancer syndrome 3; MMRCS3 [Internet]. Baltimore (MD): Johns Hopkins University. Created: Anne M. Stumpf, 13 Nov 2020 (edited 24 Nov 2020). OMIM 619097. Accessed 7 Oct 2026.
11. Online Mendelian Inheritance in Man, OMIM®. #619101 Mismatch repair cancer syndrome 4; MMRCS4 [Internet]. Baltimore (MD): Johns Hopkins University. Created: Anne M. Stumpf, 16 Nov 2020; last edited 29 Sep 2023. OMIM 619101. Accessed 7 Oct 2026.
12. GeneGuidance. Gene-specific clinical curation: MLH1, MSH2, MSH6 and PMS2 (local database, version of 26 Sep 2026). Primary references listed: Dominguez-Valentin M, et al. Genet Med. 2020 (PMID:31337882); Win AK, et al. Cancer Epidemiol Biomarkers Prev. 2017 (PMID:27799157); ClinGen Actionability AC069 and AC1055.
13. Freeman PJ, Hart RK, Gretton LJ, Brookes AJ, Dalgleish R. VariantValidator: Accurate validation, mapping, and formatting of sequence variation descriptions. Hum Mutat. 2018;39(1):61-68. doi:10.1002/humu.23348. PMID:28967166; PMC5765404. Service VariantValidator (GRCh38, transcripts MANE Select) accessed 7 Oct 2026.
14. Forster VJ, Aronson M, Zhang C, Chung J, Sudhaman S, Galati MA, et al. Biallelic EPCAM deletions induce tissue-specific DNA repair deficiency and cancer predisposition. NPJ Precis Oncol. 2024;8(1):69. doi:10.1038/s41698-024-00537-6. PMID:38467830; PMC10928233.
15. Das A, MacFarland SP, Meade J, Hansford JR, Schneider KW, Kuiper RP, et al. Clinical Updates and Surveillance Recommendations for DNA Replication Repair Deficiency Syndromes in Children and Young Adults. Clin Cancer Res. 2024;30(16):3378-3387. doi:10.1158/1078-0432.CCR-23-3994. PMID:38860976.
16. Colas C, Guerrini-Rousseau L, Suerink M, Gallon R, Kratz CP, Ayuso É, et al. ERN GENTURIS guidelines on constitutional mismatch repair deficiency diagnosis, genetic counselling, surveillance, quality of life, and clinical management. Eur J Hum Genet. 2024;32(12):1526-1541. doi:10.1038/s41431-024-01708-6. PMID:39420201; PMC11607302.
17. Idos G, Hampel H, Valle L. Lynch Syndrome. In: Adam MP, Bick S, Mirzaa GM, et al., editors. GeneReviews® [Internet]. Seattle (WA): University of Washington, Seattle; 1993–2026. Published 5 Feb 2004; comprehensive update on 17 Sep 2026. NBK1211. PMID:20301390.
18. Vasen HFA, Wimmer K, van Kouwen M, Guerrini-Rousseau L, Gattini D, Stengs L, et al. The impact of international care networks on the clinical management of constitutional mismatch repair deficiency (CMMRD): a review of recent developments. Fam Cancer. 2026;25(3). doi:10.1007/s10689-026-00584-x. PMID:42348008; PMC13303765.
19. Ercan AB, Aronson M, Fernandez NR, Chang Y, Levine A, Liu ZA, et al. Clinical and biological landscape of constitutional mismatch-repair deficiency syndrome: an International Replication Repair Deficiency Consortium cohort study. Lancet Oncol. 2024;25(5):668-682. doi:10.1016/S1470-2045(24)00026-3. PMID:38552658.
20. Aronson M, Colas C, Shuen A, Hampel H, Foulkes WD, Baris Feldman H, et al. Diagnostic criteria for constitutional mismatch repair deficiency (CMMRD): recommendations from the international consensus working group. J Med Genet. 2022;59(4):318-327. doi:10.1136/jmedgenet-2020-107627. PMID:33622763.
21. Das A, Sudhaman S, Morgenstern D, Coblentz A, Chung J, Stone SC, et al. Genomic predictors of response to PD-1 inhibition in children with germline DNA replication repair deficiency. Nat Med. 2022;28(1):125-135. doi:10.1038/s41591-021-01581-6. PMID:34992263; PMC8799468.
22. Merck Sharp & Dohme. KEYTRUDA (pembrolizumab) injection, for intravenous use. Prescribing information (BLA 125514), revised 10 Jul 2026 [Internet]. openFDA set_id 9333c79b-d487-4538-a9f0-71b91a02b287. Accessed 7 Oct 2026.
23. European Medicines Agency. Keytruda (pembrolizumab): EPAR — Product information (marketing authorisation on 17 Jul 2015; revision 72) [Internet]. EPAR. Accessed 7 Oct 2026.
24. ClinGen Actionability Working Group. Mismatch repair cancer syndrome (MMRCS): Actionability Pediatric Summary Report AC1055, release 1.0.2 [Internet]. Report. Accessed 7 Oct 2026.
25. European Reference Network on Genetic Tumour Risk Syndromes (ERN GENTURIS). Thematic disease groups; Clinical practice guidelines [Internet]. genturis.eu. Accessed 7 Oct 2026.
26. European Reference Network for Paediatric Oncology (ERN PaedCan) [Internet]. paedcan.ern-net.eu. Accessed 7 Oct 2026.
27. International Replication Repair Deficiency Consortium (IRRDC), The Hospital for Sick Children. Institutional website, “Patients and Families” and “Clinical Trials” [Internet]. replicationrepair.ca. Accessed 7 Oct 2026.
28. TURCOT J, DESPRES JP, ST PIERRE F. Malignant tumors of the central nervous system associated with familial polyposis of the colon: report of two cases. Dis Colon Rectum. 1959;2:465-8. doi:10.1007/BF02616938. PMID:13839882.
29. Hamilton SR, Liu B, Parsons RE, Papadopoulos N, Jen J, Powell SM, et al. The molecular basis of Turcot's syndrome. N Engl J Med. 1995;332(13):839-47. doi:10.1056/NEJM199503303321302. PMID:7661930.
30. Firth HV, Hurst JA. Oxford Desk Reference: Clinical Genetics and Genomics. 2nd ed. Oxford: Oxford University Press; 2017. Chapter 4, Cancer: Familial adenomatous polyposis (FAP) and adenomatous polyposis due to MUTYH, NTHL1, POLE, and POLD1. p. 569–571.
31. Ricciardone MD, Ozçelik T, Cevher B, Ozdağ H, Tuncer M, Gürgey A, et al. Human MLH1 deficiency predisposes to hematological malignancy and neurofibromatosis type 1. Cancer Res. 1999;59(2):290-3. PMID:9927033
32. Wang Q, Lasset C, Desseigne F, Frappaz D, Bergeron C, Navarro C, et al. Neurofibromatosis and early onset of cancers in hMLH1-deficient children. Cancer Res. 1999;59(2):294-7. PMID:9927034
33. Wimmer K, Etzler J. Constitutional mismatch repair-deficiency syndrome: have we so far seen only the tip of an iceberg? Hum Genet. 2008;124(2):105-22. doi:10.1007/s00439-008-0542-4. PMID:18709565.
34. Ostergaard JR, Sunde L, Okkels H. Neurofibromatosis von Recklinghausen type I phenotype and early onset of cancers in siblings compound heterozygous for mutations in MSH6. Am J Med Genet A. 2005;139A(2):96-105. doi:10.1002/ajmg.a.30998. PMID:16283678.
35. Vasen HF, Ghorbanoghli Z, Bourdeaut F, Cabaret O, Caron O, Duval A, et al. Guidelines for surveillance of individuals with constitutional mismatch repair-deficiency proposed by the European Consortium "Care for CMMR-D" (C4CMMR-D). J Med Genet. 2014;51(5):283-93. doi:10.1136/jmedgenet-2013-102238. PMID:24556086.
36. Tabori U, Hansford JR, Achatz MI, Kratz CP, Plon SE, Frebourg T, et al. Clinical Management and Tumor Surveillance Recommendations of Inherited Mismatch Repair Deficiency in Childhood. Clin Cancer Res. 2017;23(11):e32-e37. doi:10.1158/1078-0432.CCR-17-0574. PMID:28572265.
37. Durno C, Ercan AB, Bianchi V, Edwards M, Aronson M, Galati M, et al. Survival Benefit for Individuals With Constitutional Mismatch Repair Deficiency Undergoing Surveillance. J Clin Oncol. 2021;39(25):2779-2790. doi:10.1200/JCO.20.02636. PMID:33945292; PMC8407605.
38. Bouffet E, Larouche V, Campbell BB, Merico D, de Borja R, Aronson M, et al. Immune Checkpoint Inhibition for Hypermutant Glioblastoma Multiforme Resulting From Germline Biallelic Mismatch Repair Deficiency. J Clin Oncol. 2016;34(19):2206-11. doi:10.1200/JCO.2016.66.6552. PMID:27001570.
39. Suerink M, Ripperger T, Messiaen L, Menko FH, Bourdeaut F, Colas C, et al. Constitutional mismatch repair deficiency as a differential diagnosis of neurofibromatosis type 1: consensus guidelines for testing a child without malignancy. J Med Genet. 2019;56(2):53-62. doi:10.1136/jmedgenet-2018-105664. PMID:30415209.
40. Kroeze E, Weijers DD, Hagleitner MM, de Groot-Kruseman HA, Jongmans MCJ, Kuiper RP, et al. High Prevalence of Constitutional Mismatch Repair Deficiency in a Pediatric T-cell Lymphoblastic Lymphoma Cohort. Hemasphere. 2022;6(1):e668. doi:10.1097/HS9.0000000000000668. PMID:34964038; PMC8697338.
41. Negm L, Chung J, Nobre L, Bennett J, Fernandez NR, Nunes NM, et al. The landscape of primary mismatch repair deficient gliomas in children, adolescents, and young adults: a multi-cohort study. Lancet Oncol. 2025;26(1):123-135. doi:10.1016/S1470-2045(24)00640-5. PMID:39701117.
42. Ingham D, Diggle CP, Berry I, Bristow CA, Hayward BE, Rahman N, et al. Simple detection of germline microsatellite instability for diagnosis of constitutional mismatch repair cancer syndrome. Hum Mutat. 2013;34(6):847-52. doi:10.1002/humu.22311. PMID:23483711.
43. Gallon R, Phelps R, Hayes C, Brugieres L, Guerrini-Rousseau L, Colas C, et al. Constitutional Microsatellite Instability, Genotype, and Phenotype Correlations in Constitutional Mismatch Repair Deficiency. Gastroenterology. 2023;164(4):579-592.e8. doi:10.1053/j.gastro.2022.12.017. PMID:36586540.
44. Chung J, Negm L, Bianchi V, Stengs L, Das A, Liu ZA, et al. Genomic Microsatellite Signatures Identify Germline Mismatch Repair Deficiency and Risk of Cancer Onset. J Clin Oncol. 2023;41(4):766-777. doi:10.1200/JCO.21.02873. PMID:36240479; PMC10489375.
45. Chung J, Maruvka YE, Sudhaman S, Kelly J, Haradhvala NJ, Bianchi V, et al. DNA Polymerase and Mismatch Repair Exert Distinct Microsatellite Instability Signatures in Normal and Malignant Human Cells. Cancer Discov. 2021;11(5):1176-1191. doi:10.1158/2159-8290.CD-20-0790. PMID:33355208; PMC8223607.
46. Harrold E, Stadler ZK. Cancer predisposition syndromes: Lynch syndrome. In: ESMO Handbook — Cancer Genomics. Lugano: ESMO Press. p. 25–26.
47. Fernandez NR, Chang Y, Nunes NM, Dimayacyac JR, Levine A, Ringel A, et al. Patterns of hypermutation shape tumorigenesis and immunotherapy response in mismatch-repair-deficient glioma. Nat Genet. 2026;58(1):132-142. doi:10.1038/s41588-025-02420-x. PMID:41430480.
48. Guerrini-Rousseau L, Pasmant E, Muleris M, Abbou S, Adam-De-Beaumais T, Brugieres L, et al. Neurofibromatosis type 1 mosaicism in patients with constitutional mismatch repair deficiency. J Med Genet. 2024;61(2):158-162. doi:10.1136/jmg-2023-109235. PMID:37775264; PMC10850717.
49. Péron S, Metin A, Gardès P, Alyanakian MA, Sheridan E, Kratz CP, et al. Human PMS2 deficiency is associated with impaired immunoglobulin class switch recombination. J Exp Med. 2008;205(11):2465-72. doi:10.1084/jem.20080789. PMID:18824584; PMC2571921.
50. Tesch VK, IJspeert H, Raicht A, Rueda D, Dominguez-Pinilla N, Allende LM, et al. No Overt Clinical Immunodeficiency Despite Immune Biological Abnormalities in Patients With Constitutional Mismatch Repair Deficiency. Front Immunol. 2018;9:1506. doi:10.3389/fimmu.2018.01506. PMID:30013564; PMC6036136.
51. Bakry D, Aronson M, Durno C, Rimawi H, Farah R, Alharbi QK, et al. Genetic and clinical determinants of constitutional mismatch repair deficiency syndrome: report from the constitutional mismatch repair deficiency consortium. Eur J Cancer. 2014;50(5):987-96. doi:10.1016/j.ejca.2013.12.005. PMID:24440087.
52. Munteanu CV, Lighezan DL, Capcelea A, Chiriță-Emandi A, Trifa AP. Genotype-phenotype correlations in PMS2-associated constitutional mismatch repair deficiency: a systematic literature review. Oncol Rev. 2025;19:1679576. doi:10.3389/or.2025.1679576. PMID:41333974; PMC12665659.
53. Li L, Hamel N, Baker K, McGuffin MJ, Couillard M, Gologan A, et al. A homozygous PMS2 founder mutation with an attenuated constitutional mismatch repair deficiency phenotype. J Med Genet. 2015;52(5):348-52. doi:10.1136/jmedgenet-2014-102934. PMID:25691505.
54. Gallon R, Brekelmans C, Martin M, Bours V, Schamschula E, Amberger A, et al. Constitutional mismatch repair deficiency mimicking Lynch syndrome is associated with hypomorphic mismatch repair gene variants. NPJ Precis Oncol. 2024;8(1):119. doi:10.1038/s41698-024-00603-z. PMID:38789506; PMC11126593.
55. Marín F, Canet-Hermida J, Bianchi V, Chung J, Wimmer K, Foulkes W, et al. A Validated Highly Sensitive Microsatellite Instability Assay Accurately Identifies Individuals Harboring Biallelic Germline PMS2 Pathogenic Variants in Constitutional Mismatch Repair Deficiency. Clin Chem. 2024;70(5):737-746. doi:10.1093/clinchem/hvae027. PMID:38531023.
56. Perez-Valencia JA, Gallon R, Chen Y, Koch J, Keller M, Oberhuber K, et al. Constitutional mismatch repair deficiency is the diagnosis in 0.41% of pathogenic NF1/SPRED1 variant negative children suspected of sporadic neurofibromatosis type 1. Genet Med. 2020;22(12):2081-2088. doi:10.1038/s41436-020-0925-z. PMID:32773772; PMC7708300.
57. National Center for Biotechnology Information. ClinVar: VCV000009237, NM_000535.7(PMS2):c.2404C>T (p.Arg802Ter); VCV000192316, NM_000535.7(PMS2):c.2002A>G (p.Ile668Val) [Internet]. VCV000009237; VCV000192316. Accessed 7 Oct 2026.
58. De Vos M, Hayward BE, Charlton R, Taylor GR, Glaser AW, Picton S, et al. PMS2 mutations in childhood cancer. J Natl Cancer Inst. 2006;98(5):358-61. doi:10.1093/jnci/djj073. PMID:16507833.
59. Chong AL, Mejia-Garcia A, Behl S, El Haffaf Z, Chénier S, Maranda B, et al. PMS2 c.2117del (p.Lys706Serfs*19) is the Most Frequent Cancer-Associated Founder Pathogenic Variant in the French-Canadian Population of Quebec, Canada. Clin Genet. 2025;108(6):747-751. doi:10.1111/cge.14784. PMID:40546006; PMC12580481.
60. Palova H, Das A, Pokorna P, Bajciova V, Pavelka Z, Jezova M, et al. Precision immuno-oncology approach for four malignant tumors in siblings with constitutional mismatch repair deficiency syndrome. NPJ Precis Oncol. 2024;8(1):110. doi:10.1038/s41698-024-00597-8. PMID:38773265; PMC11109258.
61. Guerrini-Rousseau L, Varlet P, Colas C, Andreiuolo F, Bourdeaut F, Dahan K, et al. Constitutional mismatch repair deficiency-associated brain tumors: report from the European C4CMMRD consortium. Neurooncol Adv. 2019;1(1):vdz033. doi:10.1093/noajnl/vdz033. PMID:32642664; PMC7212899.
62. Raveneau M, Guerrini-Rousseau L, Levy R, Roux CJ, Bolle S, Doz F, et al. Specific brain MRI features of constitutional mismatch repair deficiency syndrome in children with high-grade gliomas. Eur Radiol. 2024;34(12):7765-7775. doi:10.1007/s00330-024-10885-3. PMID:38981890.
63. Suwala AK, Friedel D, Hinz FE, Mahlknecht PD, Schinkewitsch S, Rieder M, et al. Primary mismatch repair deficient glioma, IDH-wildtype and H3-wildtype: A giant cell tumor with potential for long-term survival occurring at all ages. Neuro Oncol. 2026;28(9):2153-2165. doi:10.1093/neuonc/noag132. PMID:42236272; PMC13550676.
64. Ripperger T, Schlegelberger B. Acute lymphoblastic leukemia and lymphoma in the context of constitutional mismatch repair deficiency syndrome. Eur J Med Genet. 2016;59(3):133-42. doi:10.1016/j.ejmg.2015.12.014. PMID:26743104.
65. Rigaud C, Forster VJ, Al-Tarrah H, Attarbaschi A, Bianchi V, Burke A, et al. Comprehensive analysis of constitutional mismatch repair deficiency-associated non-Hodgkin lymphomas in a global cohort. Pediatr Blood Cancer. 2024;71(12):e31302. doi:10.1002/pbc.31302. PMID:39300701.
66. Levi Z, Kariv R, Barnes-Kedar I, Goldberg Y, Half E, Morgentern S, et al. The gastrointestinal manifestation of constitutional mismatch repair deficiency syndrome: from a single adenoma to polyposis-like phenotype and early onset cancer. Clin Genet. 2015;88(5):474-8. doi:10.1111/cge.12518. PMID:25307252.
67. Kratz CP, Holter S, Etzler J, Lauten M, Pollett A, Niemeyer CM, et al. Rhabdomyosarcoma in patients with constitutional mismatch-repair-deficiency syndrome. J Med Genet. 2009;46(6):418-20. doi:10.1136/jmg.2008.064212. PMID:19293170.
68. Shiran SI, Ben-Sira L, Elhasid R, Roth J, Tabori U, Yalon M, et al. Multiple Brain Developmental Venous Anomalies as a Marker for Constitutional Mismatch Repair Deficiency Syndrome. AJNR Am J Neuroradiol. 2018;39(10):1943-1946. doi:10.3174/ajnr.A5766. PMID:30166433; PMC7410739.
69. Kerpel A, Yalon M, Soudack M, Chiang J, Gajjar A, Nichols KE, et al. Neuroimaging Findings in Children with Constitutional Mismatch Repair Deficiency Syndrome. AJNR Am J Neuroradiol. 2020;41(5):904-910. doi:10.3174/ajnr.A6512. PMID:32354708; PMC7228156.
70. Baas AF, Gabbett M, Rimac M, Kansikas M, Raphael M, Nievelstein RA, et al. Agenesis of the corpus callosum and gray matter heterotopia in three patients with constitutional mismatch repair deficiency syndrome. Eur J Hum Genet. 2013;21(1):55-61. doi:10.1038/ejhg.2012.117. PMID:22692065; PMC3522206.
71. Toledano H, Orenstein N, Sofrin E, Ruhrman-Shahar N, Amarilyo G, Basel-Salmon L, et al. Paediatric systemic lupus erythematosus as a manifestation of constitutional mismatch repair deficiency. J Med Genet. 2020;57(7):505-508. doi:10.1136/jmedgenet-2019-106303. PMID:31501241.
72. Firth HV, Hurst JA. Oxford Desk Reference: Clinical Genetics and Genomics. 2nd ed. Oxford: Oxford University Press; 2017. Chapter 4, Cancer: Colorectal cancer (CRC). p. 556–557.
73. Firth HV, Hurst JA. Oxford Desk Reference: Clinical Genetics and Genomics. 2nd ed. Oxford: Oxford University Press; 2017. Chapter 2, Clinical approach: Patchy pigmented skin lesions (café-au-lait). p. 278–279.
74. Firth HV, Hurst JA. Oxford Desk Reference: Clinical Genetics and Genomics. 2nd ed. Oxford: Oxford University Press; 2017. Chapter 3, Common consultations: Neurofibromatosis type 1 (NF1). p. 506–511.
75. Suerink M, Potjer TP, Versluijs AB, Ten Broeke SW, Tops CM, Wimmer K, et al. Constitutional mismatch repair deficiency in a healthy child: On the spot diagnosis? Clin Genet. 2018;93(1):134-137. doi:10.1111/cge.13053. PMID:28503822.
76. Friker LL, Perwein T, Waha A, Dörner E, Klein R, Blattner-Johnson M, et al. MSH2, MSH6, MLH1, and PMS2 immunohistochemistry as highly sensitive screening method for DNA mismatch repair deficiency syndromes in pediatric high-grade glioma. Acta Neuropathol. 2025;149(1):11. doi:10.1007/s00401-025-02846-x. PMID:39894875; PMC11788232.
77. Giunti L, Cetica V, Ricci U, Giglio S, Sardi I, Paglierani M, et al. Type A microsatellite instability in pediatric gliomas as an indicator of Turcot syndrome. Eur J Hum Genet. 2009;17(7):919-27. doi:10.1038/ejhg.2008.271. PMID:19156169; PMC2986495.
78. Sehested A, Meade J, Scheie D, Østrup O, Bertelsen B, Misiakou MA, et al. Constitutional POLE variants causing a phenotype reminiscent of constitutional mismatch repair deficiency. Hum Mutat. 2022;43(1):85-96. doi:10.1002/humu.24299. PMID:34816535.
79. van der Klift HM, Mensenkamp AR, Drost M, Bik EC, Vos YJ, Gille HJ, et al. Comprehensive Mutation Analysis of PMS2 in a Large Cohort of Probands Suspected of Lynch Syndrome or Constitutional Mismatch Repair Deficiency Syndrome. Hum Mutat. 2016;37(11):1162-1179. doi:10.1002/humu.23052. PMID:27435373.
80. Shuen AY, Lanni S, Panigrahi GB, Edwards M, Yu L, Campbell BB, et al. Functional Repair Assay for the Diagnosis of Constitutional Mismatch Repair Deficiency From Non-Neoplastic Tissue. J Clin Oncol. 2019;37(6):461-470. doi:10.1200/JCO.18.00474. PMID:30608896.
81. Gallon R, Mühlegger B, Wenzel SS, Sheth H, Hayes C, Aretz S, et al. A sensitive and scalable microsatellite instability assay to diagnose constitutional mismatch repair deficiency by sequencing of peripheral blood leukocytes. Hum Mutat. 2019;40(5):649-655. doi:10.1002/humu.23721. PMID:30740824; PMC6519362.
82. González-Acosta M, Marín F, Puliafito B, Bonifaci N, Fernández A, Navarro M, et al. High-sensitivity microsatellite instability assessment for the detection of mismatch repair defects in normal tissue of biallelic germline mismatch repair mutation carriers. J Med Genet. 2020;57(4):269-273. doi:10.1136/jmedgenet-2019-106272. PMID:31494577; PMC7146943.
83. Bodo S, Colas C, Buhard O, Collura A, Tinat J, Lavoine N, et al. Diagnosis of Constitutional Mismatch Repair-Deficiency Syndrome Based on Microsatellite Instability and Lymphocyte Tolerance to Methylating Agents. Gastroenterology. 2015;149(4):1017-29.e3. doi:10.1053/j.gastro.2015.06.013. PMID:26116798.
84. Firth HV, Hurst JA. Oxford Desk Reference: Clinical Genetics and Genomics. 2nd ed. Oxford: Oxford University Press; 2017. Chapter 4, Cancer: Juvenile polyposis syndrome (JPS). p. 578–580.
85. Wimmer K, Rosenbaum T, Messiaen L. Connections between constitutional mismatch repair deficiency syndrome and neurofibromatosis type 1. Clin Genet. 2017;91(4):507-519. doi:10.1111/cge.12904. PMID:27779754.
86. Wimmer K, Beilken A, Nustede R, Ripperger T, Lamottke B, Ure B, et al. A novel germline POLE mutation causes an early onset cancer prone syndrome mimicking constitutional mismatch repair deficiency. Fam Cancer. 2017;16(1):67-71. doi:10.1007/s10689-016-9925-1. PMID:27573199; PMC5243902.
87. Bessis D, Palenzuela G, Roujeau T, Willems M. Germline POLE variant with constitutional mismatch-repair deficiency-like skin features and associated high-grade gliomas in a child. Ann Dermatol Venereol. 2026;153(3):103517. doi:10.1016/j.annder.2026.103517. PMID:42468069.
88. Itoh H, Ohsato K. Turcot syndrome and its characteristic colonic manifestations. Dis Colon Rectum. 1985;28(6):399-402. doi:10.1007/BF02560221. PMID:4006635.
89. Paraf F, Jothy S, Van Meir EG. Brain tumor-polyposis syndrome: two genetic diseases? J Clin Oncol. 1997;15(7):2744-58. doi:10.1200/JCO.1997.15.7.2744. PMID:9215849.
90. Ghorbanoghli Z, van Kouwen M, Versluys B, Bonnet D, Devalck C, Tinat J, et al. High yield of surveillance in patients diagnosed with constitutional mismatch repair deficiency. J Med Genet. 2023;60(7):679-684. doi:10.1136/jmg-2022-108829. PMID:36411031.
91. Lavoine N, Colas C, Muleris M, Bodo S, Duval A, Entz-Werle N, et al. Constitutional mismatch repair deficiency syndrome: clinical description in a French cohort. J Med Genet. 2015;52(11):770-8. doi:10.1136/jmedgenet-2015-103299. PMID:26318770.
92. ESMO Handbook — Cancer Genomics. Lugano: ESMO Press. Genomics in gastrointestinal cancers: colorectal cancer, advanced disease. p. 55.
93. Selesner L, Vaughn CR, Jazrawi S, Fenlon E, Storm K, Moyer K, et al. Immunotherapy in Pediatric Constitutional Mismatch Repair Deficiency (CMMRD)-Associated Colorectal Cancer: Report of a Rare Variant and Recommendations for Screening. J Pediatr Hematol Oncol. 2026;48(6):e273-e276. doi:10.1097/MPH.0000000000003228. PMID:42330152.
94. Das A, Tabori U, Sambira Nahum LC, Collins NB, Deyell R, Dvir R, et al. Efficacy of Nivolumab in Pediatric Cancers with High Mutation Burden and Mismatch Repair Deficiency. Clin Cancer Res. 2023;29(23):4770-4783. doi:10.1158/1078-0432.CCR-23-0411. PMID:37126021; PMC10690097.
95. The Hospital for Sick Children. Pilot Study of Nivolumab in Pediatric Patients With Hypermutant Cancers. ClinicalTrials.gov: NCT02992964. Terminated; last updated 2 Apr 2024. Accessed 7 Oct 2026.
96. Das A, Fernandez NR, Levine A, Bianchi V, Stengs LK, Chung J, et al. Combined Immunotherapy Improves Outcome for Replication-Repair-Deficient (RRD) High-Grade Glioma Failing Anti-PD-1 Monotherapy: A Report from the International RRD Consortium. Cancer Discov. 2024;14(2):258-273. doi:10.1158/2159-8290.CD-23-0559. PMID:37823831; PMC10850948.
97. Marcus L, Lemery SJ, Keegan P, Pazdur R. FDA Approval Summary: Pembrolizumab for the Treatment of Microsatellite Instability-High Solid Tumors. Clin Cancer Res. 2019;25(13):3753-3758. doi:10.1158/1078-0432.CCR-18-4070. PMID:30787022.
98. U.S. Food and Drug Administration. Supplement approval letter BLA 125514/S-014 (KEYTRUDA, MSI-H/dMMR solid tumors; accelerated approval), 23 May 2017 [Internet]. Letter. Accessed 7 Oct 2026.
99. U.S. Food and Drug Administration. Supplement approval letter BLA 125554/S-034 (OPDIVO, MSI-H/dMMR metastatic colorectal cancer), 31 Jul 2017 [Internet]. Letter. Accessed 7 Oct 2026.
100. U.S. Food and Drug Administration. Supplement approval letter BLA 125554/S-063 (OPDIVO in combination with ipilimumab, MSI-H/dMMR metastatic colorectal cancer), 10 Jul 2018 [Internet]. Letter. Accessed 7 Oct 2026.
101. Bristol-Myers Squibb. OPDIVO (nivolumab) injection, for intravenous use. Prescribing information (BLA 125554), revised 12 Aug 2026 [Internet]. openFDA set_id f570b9c4-6846-4de2-abfa-4d0a4ae4e394. Accessed 7 Oct 2026.
102. European Medicines Agency. Opdivo (nivolumab): EPAR — Product information (authorisation on 19 Jun 2015; revision 76); Yervoy (ipilimumab): authorisation on 13 Jul 2011 [Internet]. EPAR Opdivo. Accessed 7 Oct 2026.
103. GlaxoSmithKline. JEMPERLI (dostarlimab-gxly) injection. Prescribing information (BLA 761174), revised 2 Sep 2025 [Internet]. openFDA set_id 095eab9f-545a-4f12-bfb7-19477fb901a5. Accessed 7 Oct 2026.
104. European Medicines Agency. Jemperli (dostarlimab): EPAR — Product information (authorisation on 21 Apr 2021) [Internet]. EPAR. Accessed 7 Oct 2026.
105. Bristol-Myers Squibb. YERVOY (ipilimumab) injection, for intravenous use. Prescribing information; label version dated 20 Aug 2026 [Internet]. openFDA. Accessed 26 Oct 2026.
106. European Medicines Agency. Yervoy (ipilimumab): EPAR — product information (summary of product characteristics, sections 4.1 and 4.2) [Internet]. PDF. Accessed 26 Oct 2026.
107. Novartis. MEKINIST (trametinib) tablets and for oral solution. Prescribing information; label version dated 7 May 2026 [Internet]. openFDA. Accessed 26 Oct 2026.
108. European Medicines Agency. Mekinist (trametinib): EPAR — product information (summary of product characteristics, section 4.2) [Internet]. PDF. Accessed 26 Oct 2026.
109. Leenders EKSM, Westdorp H, Brüggemann RJ, Loeffen J, Kratz C, Burn J, et al. Cancer prevention by aspirin in children with Constitutional Mismatch Repair Deficiency (CMMRD). Eur J Hum Genet. 2018;26(10):1417-1423. doi:10.1038/s41431-018-0197-0. PMID:29904176; PMC6138701.
110. Bayer Portugal. Aspirina GR 100 mg comprimidos gastrorresistentes (ácido acetilsalicílico). Summary of product characteristics [Internet]. INFARMED, Infomed. SmPC. Accessed 26 Oct 2026.
111. Kloor M, Reuschenbach M, Pauligk C, Karbach J, Rafiyan MR, Al-Batran SE, et al. A Frameshift Peptide Neoantigen-Based Vaccine for Mismatch Repair-Deficient Cancers: A Phase I/IIa Clinical Trial. Clin Cancer Res. 2020;26(17):4503-4510. doi:10.1158/1078-0432.CCR-19-3517. PMID:32540851.
112. Morgenstern D (sponsor). U-R-Immune Glioma (CA209-1245): upfront adaptive immunotherapy in CAYA patients with replication-repair deficient high-grade gliomas. ClinicalTrials.gov: NCT06519682. Recruiting; last updated 29 Aug 2025. Accessed 7 Oct 2026.
113. Nationwide Children's Hospital. Feasibility and Phase 2a Study of Relatlimab and Nivolumab (Opdualag) in Adolescent and Young Adult Patients Newly Diagnosed With Replication Repair Deficient High-Grade Glioma, Including DIPG. ClinicalTrials.gov: NCT07644312. Not yet recruiting; first posted 12 Jun 2026; last updated 9 Sep 2026. Accessed 7 Oct 2026.
114. Cancer Research UK. DETERMINE Trial Treatment Arm 02: Atezolizumab in Adult, Paediatric and Teenage/Young Adult Patients With Cancers With High TMB or MSI-high or Proven CMMRD Disposition. ClinicalTrials.gov: NCT05770102. Recruiting; last updated 24 Nov 2025. Accessed 7 Oct 2026.
115. National Cancer Institute. Pembrolizumab in Treating Younger Patients With Recurrent, Progressive, or Refractory High-Grade Gliomas, DIPG, Ependymoma, Medulloblastoma or Hypermutated Brain Tumors. ClinicalTrials.gov: NCT02359565. Active, not recruiting; last updated 20 Jul 2026. Accessed 7 Oct 2026.
116. National Cancer Institute. 3CI Study: Childhood Cancer Combination Immunotherapy — nivolumab and ipilimumab in CAYA patients with relapsed/refractory hypermutant cancers. ClinicalTrials.gov: NCT04500548. Withdrawn (insufficient recruitment); last updated 10 Jun 2026. Accessed 7 Oct 2026.
117. Nair G, Cadenas N, Singh A. A missed opportunity for preventing CMMRD: is it time to include Lynch syndrome genes in prenatal genetic testing? BMJ Case Rep. 2025;18(11). doi:10.1136/bcr-2025-267499. PMID:41224689; PMC12625528.
118. Farschtschi SC, Kumps C, Milagre TH, Makrythanasis P, Van Tongerloo A, Denayer E, et al. ERN GENTURIS guideline on counselling on reproductive options for individuals with a cancer predisposition syndrome (including genturis). Eur J Hum Genet. 2026;34(3):307-313. doi:10.1038/s41431-025-02007-4. PMID:41530368; PMC12963376.
119. Amayiri N, Tabori U, Campbell B, Bakry D, Aronson M, Durno C, et al. High frequency of mismatch repair deficiency among pediatric high grade gliomas in Jordan. Int J Cancer. 2016;138(2):380-5. doi:10.1002/ijc.29724. PMID:26293621.
120. Mushtaq N, Minhas K, Bashir F, Zahid S, Qureshi BM, Javed G, et al. Frequency and Impact of Constitutional Mismatch Repair Deficiency in Patients With High-Grade Glioma, a Retrospective Analysis of 7 Years in Pakistan: an IRRDC Study. JCO Glob Oncol. 2025;11:e2400247. doi:10.1200/GO-24-00247. PMID:39819126.
121. Cohen S, Cozacov T, Weintraub I, Shaoul R, Ledder O, Yogev D, et al. Clinical and genetic features of pediatric hereditary polyposis syndromes in Israel: A nationwide multicenter cohort. J Pediatr Gastroenterol Nutr. 2026. Advance electronic publication. doi:10.1002/jpn3.70500. PMID:42439066.
122. Baris HN, Barnes-Kedar I, Toledano H, Halpern M, Hershkovitz D, Lossos A, et al. Constitutional Mismatch Repair Deficiency in Israel: High Proportion of Founder Mutations in MMR Genes and Consanguinity. Pediatr Blood Cancer. 2016;63(3):418-27. doi:10.1002/pbc.25818. PMID:26544533.
123. Ahmad O, Sultan I, Hussaini M, Alsmadi O, Sutter C, Hirsch S, et al. Molecular characterization of mismatch repair deficient tumors in young Jordanian patients. Discov Oncol. 2026;17(1). doi:10.1007/s12672-026-05717-3. PMID:42608506; PMC13481598.