Article Text

Download PDFPDF

hMLH1 and hMSH2 immunostaining in colorectal cancer
  1. J R JASS
  1. Department of Pathology, University of Queensland
  2. Medical School, Herston, Qld 4006, Australia
  3. Email: j.jass{at}
  2. M F DIXON,
  1. Molecular Oncology
  2. Algernon Firth Institute of Pathology
  3. School of Medicine, University of Leeds
  4. Leeds LS2 9JT, UK
  5. Email: L.Cawkwell{at}
  1. Academic Department of Pathology
  2. St Mark's Hospital, Northwick Park
  3. Watford Road, Harrow HA1 3UJ, UK
  4. Email: i.talbot{at}

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

Editor,—The paper by Cawkwell and colleagues (OpenUrlAbstract/FREE Full Text) on the utility of hMLH1 and hMSH2 immunostaining in colorectal cancer may mislead the unwary reader just as it misled the author of the accompanying commentary (OpenUrlAbstract/FREE Full Text) . Nowhere in their paper do the authors state that their approach will identify all cases of hereditary non-polyposis colorectal cancer (HNPCC). Nor would their method of ascertainment have picked up many HNPCC families. This is evident from the high proportion (83%) of cases with loss of hMLH1 while only four (apparently) of the 49 subjects with one or more RER positive colorectal carcinomas were diagnosed at less than 50 years of age. No subject was actually confirmed as having HNPCC. Yet the commentary states that the test showed that all HNPCC subjects had a deficit of either hMLH1 or hMSH2.

The test will certainly identify all sporadic RER positive or microsatellite instability-high (MSI-H) colorectal cancers in which the promoter region of hMLH1 is hypermethylated.1 We found that 21/23 previously reported sporadic MSI-H cancers2showed loss of hMLH1. One showed loss of hMSH2. This subject was adopted as a child and developed his cancer at the age of 34 years. He probably had HNPCC. The other cancer retaining both hMLH1 and hMSH2 was on the borderline of MSI-L and MSI-H and had probably been assigned as MSI-H incorrectly. In contrast, none of 41 microsatellite stable nor 19 microsatellite-low (MSI-L) cases showed loss of hMLH1 or hMSH2.

The immunohistochemical approach will identify some but not all HNPCC cancers. The issues are as follows:

Genes other than hMLH1 and hMSH2 cause HNPCC.3
Subtly mutated proteins may retain antigenicity while losing function.3
Cancers in some HNPCC subjects may retain DNA repair proficiency.4
Not all HNPCC kindreds develop colorectal cancer.
Antigen retrieval may be technically difficult in old tissue blocks.4

It is essential that these caveats be understood before there is a major change in management strategy. A wider net is required to pick up all HNPCC families and this includes both routine morphological assessment and testing for DNA microsatellite instability as well as obtaining a family history on all subjects with colorectal cancer. Morphological assessment is not specific but costs little and will identify over 90% of HNPCC cancers regardless of microsatellite status or mismatch repair protein expression status (unpublished observations).

Notwithstanding the words of caution, immunohistochemistry will serve as a major advance in the work up of suspected HNPCC families. It is particularly useful in identifying the underlying germline mutation and thereby facilitating genetic testing. The impact of the test on sporadic case management can certainly be anticipated but warrants further evaluation.



Editor,—We are pleased that Professor Jass believes that immunohistochemistry will serve as a major advance in the work up of families with suspected hereditary non-polyposis colorectal cancer (HNPCC). We are also in absolute agreement that our immunohistochemical test is unlikely to detect all cases of true HNPCC. Our paper makes no claims to the contrary. The cases used in our study were subgrouped according to simple criteria such as patient age, and location and multiplicity of carcinomas. Our study design did not include a series of known HNPCC carcinomas and therefore we could not, and did not attempt to, state the value of the test in detecting true HNPCC carcinomas. Our main finding in the paper was the potential value in the sporadic setting of routinely staining all colorectal carcinomas using antibodies against hMSH2 and hMLH1. This would give information on prognosis, risk of metachronous colorectal carcinomas, and identify a group of patients who should be investigated further for HNPCC. However, the majority of cancers which exhibited loss of expression of the hMLH1 protein in our study are likely to be sporadic cases with hypermethylation of the hMLH1 promoter. It certainly would be important to assess the value of immunohistochemistry for the detection of true HNPCC cases but a large well characterised series with definite family history and known germline and somatic defects would ideally need to be assembled to fully answer this question. We have early data which suggest that the antibodies may have an important role and this is in preparation for submission as a paper.

The question of successful antigen retrieval from old tissue blocks is valid, but we have not encountered significant problems in further clinical series of 400 cases in the AXIS trial and 400 cases in the QUASAR 1 trial. Our paper mainly suggests the prospective assessment of all cases of colorectal cancer as they are diagnosed, therefore utilising new blocks.


Editor,—I feel that Professor Jass misrepresents what I said in my commentary. I drew attention to the value of the method for detecting expression of the mismatch repair proteins MLH1 and MSH2 as a screening method for tumours showing deficient expression of one or other of these two proteins. I pointed out that there had not been a case of hereditary non-polyposis colorectal cancer (HNPCC) proved to be due to genetic loss of any other gene although there was a theoretical possibility that other genes could be involved. I did not state that the (immunohistochemical) test showed that all HNPCC subjects had a deficit of one of these two genes and I certainly did not say that by immunohistochemical staining for MLH1 and MSH2, all cases of HNPCC would be identified.

Jass has usefully widened the discussion about mismatch repair gene deficient colorectal cancers by drawing attention to his own validation of the morphological features of RER positive colorectal cancers.2-1 2-2 I agree that it is entirely possible that the immunohistochemical method could fail to identify every case of HNPCC, but it does have some degree of objectivity for detecting tumours with failed mismatch repair gene expression. Clearly, careful thought would need to be given to how this and other investigatory approaches should be applied if guidelines were to be produced for a national screening programme for HNPCC but the greater value of the immunological test may be in relation to management of patients with RER positive colorectal cancers.


  1. 2-1.
  2. 2-2.