How can we best detect hereditary non-polyposis colorectal cancer?

Although a strong genetic predisposition to colorectal cancer (CRC) is rare, it is important because of its large contribution to CRC diagnosed before the age of 50 years and because mortality from CRC can be reduced by appropriate management. There are currently limitations in determining whether a case of CRC is “sporadic” or whether it might be related to an inherited predisposition. However, Australian research that examines the utility of new diagnostic tools to help diagnose genetic tendency to CRC may be showing us a way forward.1,2

The two best-characterised high-risk inherited syndromes are familial adenomatous polyposis and hereditary non-polyposis colorectal cancer (HNPCC) — sometimes known as Lynch syndrome. Both are inherited as autosomal dominant traits.

Familial adenomatous polyposis can usually be readily diagnosed on the basis of clinical findings alone, when an individual develops CRC at a relatively young age on a background of colorectal adenomatous polyposis (mostly with more than 100 adenomas). HNPCC cannot usually be readily diagnosed.

HNPCC accounts for about 1%–4% of all CRC, but up to 10% in patients younger than 50 years at diagnosis. It is caused by a germline (heritable) mutation in one of a family of genes known as the DNA mismatch repair genes: hMLH1, hMSH2, hMSH6 and hPMS2.3 HNPCC is characterised clinically by an early age of onset of CRC, a predisposition for proximal colonic cancers, and a tendency to develop multiple CRC, along with an increased risk of some extra-colonic malignancies, including cancers of the uterus and ovary, stomach, small bowel, biliary tree, ureter, renal pelvis, pancreas and brain. The family history is often complex.

Traditionally, HNPCC has been suspected in families where the family history of cancer meets the modified Amsterdam criteria.4 These require a family to have at least three close relatives in two generations diagnosed with cancer of the colon, rectum, endometrium, small bowel, ureter or renal pelvis, with at least one diagnosed before age 50 years.

One problem is that not all “Amsterdam positive” families will be proven to have HNPCC and, conversely, some families with proven HNPCC do not meet these criteria. Family history alone is not enough — a diagnosis of “suspected HNPCC” may be based on verified clinical and pathological information from the family pedigree, but the diagnosis of HNPCC is ultimately confirmed by the demonstration of a family germline mutation in one of the mismatch repair genes. This is done by taking blood from one of the affected family members and searching the mismatch repair genes to determine whether a causative mutation can be found.

However, a mutation cannot be found in every family, as mutations may be missed or mutations may be present in other genes that are not yet identified. This means that if the family history is strong (Amsterdam positive) and the genetic test (mutation search) fails to identify a mutation in an affected family member, the test result should be considered “inconclusive” and all relatives remain at potentially high risk.

Only when a causative mutation in one of the mismatch repair genes has been identified can other at-risk adult family members be offered “predictive” genetic testing to determine their risk. Importantly, those found not to carry the family mutation should be considered at the average population risk for cancer, and they (and their offspring) can be spared additional cancer screening and concern.

At present, many families with a family history that meets or approaches the Amsterdam criteria are offered this expensive approach to germline genetic testing; few are found to have proven HNPCC.

The accurate detection of HNPCC has improved recently for a number of reasons. The first is an increasing awareness of family history as a risk factor for CRC. Clinical practice guidelines assist in estimating CRC risk based on family history.5,6 For those with a more complex family history, referral to a family cancer clinic may be appropriate. These clinics have been developed to provide cancer risk assessment, surveillance, and prevention strategies, with genetic counselling and testing when appropriate. Finally, we now have improved diagnostic tools, including molecular testing of tumours for microsatellite instability and immunohistochemical staining of tumour tissue for mismatch repair proteins, which can be used to help investigate a history of “suspected HNPCC”. Additional molecular tumour tests are in development.

Southey et al have recently conducted an analysis of the use of tumour testing to prioritise mismatch repair germline testing in an Australian population-based study of early onset (< 45 years old) CRC.1 In 131 patients, unselected for family history, they found 18 with germline mutations in one of the mismatch repair genes (indicating HNPCC). Based on family history alone (the Amsterdam criteria), only half of these would have been identified, indicating that tumour testing (specifically immunohistochemical staining) is a very useful adjunct to diagnosis, and more sensitive than family history. Ward et al, also Australian, have since published similar findings.2

Immunohistochemical staining has an advantage over microsatellite instability testing in that it is relatively inexpensive, can be conducted in a routine pathology laboratory (rather than a molecular laboratory), and the immunohistochemical staining test identifies the specific mismatch repair gene in which to search for a germline mutation, thereby saving money by directing the mutation search. However, although immunohistochemical staining is promising as a tool to increase diagnostic accuracy for HNPCC, it may be too early to apply this test routinely at diagnosis to all patients with CRC.

There are still some problems with tumour testing. Some patients have loss of staining on immunohistochemical staining without a demonstrable mismatch repair germline mutation, indicating a lack of specificity, although this may change as germline testing improves. Immunohistochemical staining utilises antibodies that are not commonly used in pathology laboratories, and staining may be patchy and difficult to interpret for those who do not use these antibodies routinely. Moreover, there is generally a lack of understanding concerning the interpretation of loss of stain — loss of hMLH1 in a patient with an older onset CRC, without family history, usually indicates a sporadic cancer rather than HNPCC, but some of these families are now inappropriately being referred to family cancer clinics as “possible HNPCC”. Finally, some argue that tumour testing is a surrogate genetic test and so requires genetic counselling and fully informed consent. However, tumour testing should be simply viewed as a useful adjunct to HNPCC diagnosis, in that an abnormal finding on immunohistochemical staining raises the possibility, rather than the certainty, of a familial predisposition to CRC.

Tumour testing could now be used as a “triage” measure to assist referral to a family cancer clinic for further assessment and germline testing, if appropriate. It is especially applicable when the family history approaches, but does not quite meet, the Amsterdam criteria for suspecting HNPCC or when CRC is diagnosed at an early age (younger than 50 years). However, before this can become routine, immunohistochemical staining testing for HNPCC may need to be standardised, and laboratories validated through the Royal College of Pathologists of Australia to ensure reliable results.

Even so, for now, the best diagnostic precision is still achieved by a combined analysis of all the variables: family history, and clinical and pathological findings, as well as results of genetic tests.