Absence of PMS2 Mutations in Colon-CFR Participants whose Colorectal Cancers Demonstrate Unexplained Loss of MLH1 Expression

To the Editor:

Lynch syndrome (LS) (MIM#276300) is a cancer predisposition condition resulting from mutations within the DNA mismatch repair (MMR) genes (MLH1, MSH2, MSH6 or PMS2). In a suspected LS family, tumors are assessed for microsatellite instability (MSI) and/or immunohistochemical (IHC) absence of MMR proteins, both of which are hallmarks of LS. To enable appropriate diagnosis, counseling and surveillance, the gene indicated by the IHC profile is screened for mutations via techniques such as denaturing high pressure liquid chromatography (DHPLC), sequencing and multiplex ligation-dependent probe amplification (MLPA). Using these approaches, mutations are identified in the majority of cases suspected of LS; however, there remain a substantial proportion of families for which mutations cannot be identified.

Tumors with loss of both MLH1 and PMS2 expression are associated with mutations in MLH1, whereas mutations in PMS2 are associated with solitary loss of PMS2. Contrary to conventional IHC indications, Niessen et al. (2009) [1] identified a PMS2 mutation (p.Q233X) in an individual whose tumor showed loss of expression of MLH1, suggesting that such mutations may explain some of these suspected LS cases. Evidence for this would necessitate a change to routine screening practices which currently have little justification to extend mutation detection beyond the primary indicated MMR gene.

We screened a cohort of individuals from the Colorectal Cancer Family Registry (C-CFR; http://epi.grants.cancer.gov/CFR) whose tumors showed loss of expression of MLH1 (and PMS2), but who had not had a genetic or epigenetic MLH1 mutation identified. The C-CFR comprises 435 probands whose tumors demonstrate IHC loss of MLH1, with a causative mutation/methylation being identified in 339 (78%) cases. DNA was available for 84 of the outlying samples, which comprised the study cohort.

Forty-nine of these samples had previously only been screened for MLH1 mutations via DHPLC. Subsequent sequencing analysis identified MLH1 mutations in four individuals (4/49, 8.2%) who were removed from the rest of the study. The remaining 80 individuals had a mean age of diagnosis of 47yrs, and were primarily from population-based recruitment (68/80, 85%). A history of cancer fulfilling established clinical criteria [2] was reported for 76 of the 80 probands (95%).

Complete sequencing and MLPA (P008-A1) analysis (described in [3]) of the PMS2 gene in these individuals identified common polymorphisms, four rare/novel variants but no truncating mutations (Table 1). The results from this study, the only one to date that has focused on this select group of individuals, found no evidence for deleterious mutations in PMS2. As such, these findings are sufficient to exclude PMS2 as a gene for mutation testing in individuals with suspected LS based on tumoral loss of both MLH1 and PMS2 expression. This conclusion is supported by earlier studies that screened moderate numbers (n=25) of LS suspected patients showing loss of both PMS2 and MLH1, [4, 5] again with no evidence of deleterious mutations in PMS2.

Table 1.

Summary of the PMS2 variants identified among the 80 individuals tested

Variant	Wild Type	Heterozygous	Homozygous	Allele Frequency in cases	Variant Type
c.59G>A p.Arg20Gln	69	6	5	10.0%	Polymorphism
c.780G>C p.Ser260Ser	57	20	3	16.3%	Polymorphism
c.983A>G p.Asp328Gly	79	1	0	0.6%	Novel
c.1364C>T p.Ser455Phe	79	1	0	0.6%	Novel
c.1408C>T p.Pro470Ser	19	45	16	48.1%	Polymorphism
c.1454C>A p.Thr485Lys	69	11	0	6.9%	Polymorphism
c.1531A>G p.Thr511Ala	77	3	0	1.9%	Polymorphism
c.1569C>G p.Ser523Ser	79	1	0	0.6%	Novel
c.1621G>A p.Glu541Lys	62	16	2	12.5%	Polymorphism
c.1688G>T p.Arg563Leu	79	1	0	0.6%	Rare
c.1789A>T p.Thr597Ser	76	4	0	2.5%	Polymorphism
c.1866G>A p.Met622Ile	79	1	0	0.6%	Polymorphism
c.2006+6G>A	74	6	0	3.8%	Polymorphism
c.2007−7C>T	69	10	1	7.5%	Polymorphism
c.2007−4G>A	53	24	3	18.8%	Polymorphism
c.2253T>C p.Phe751Phe	67	11	2	9.4%	Polymorphism
c.2324A>G p.Asn775Ser	64	13	3	11.9%	Polymorphism
c.2340C>T p.Pro780Pro	69	8	3	8.8%	Polymorphism
c.2466T>C p.Leu822Leu	41	34	5	27.5%	Polymorphism
c.2570G>C p.Gly857Ala	64	16	0	10.0%	Polymorphism
c.*17G>C	71	6	3	7.5%	Polymorphism
c.*92dupA	34	39	7	33.1%	Polymorphism

The heterodimeric nature of the MMR proteins has led to the well established practice of utilizing IHC to direct mutation screening. As such, the identification of a single PMS2 mutation carrier with loss of expression of MLH1 should not give strong reason to screen PMS2 in cases with concomitant loss of MLH1 and PMS2 in the tumor when routine screening fails to identify a mutation in MLH1. [1] It is far more logical to expect either an overlooked mutation (our finding that DHPLC can miss mutations in the MLH1 gene highlights one example of how this may occur) within the coding region of MLH1 or a more complicated mutational mechanism that would not be detected by routine screening practices. [6–8]

In conclusion, we have found no evidence to support screening of the PMS2 gene for mutations in CRC cases that demonstrate unexplained loss of MLH1 and PMS2. Therefore, further investigation of novel mutational mechanisms in MLH1 is warranted.