The paper addressing the role of monogenic inheritance in the aetiology of colorectal cancer highlights the importance of achieving a meaningful working diagnosis of hereditary non-polyposis colorectal cancer (HNPCC) (Gut 2004;53:115–22). The diagnosis of HNPCC may be achieved in two ways:

1. by equating the Amsterdam criteria with a clinical diagnosis,

2. by compiling a comprehensive set of clinical, pathological, and molecular features that would together support the diagnosis of a specific condition caused by a germline mutation in a DNA mismatch repair gene such as hMLH1 or hMSH2.

It is clear that there is potential for considerable confusion if the label HNPCC is used in these very different ways.

The original Amsterdam criteria were not developed to serve as the diagnostic criteria for HNPCC but merely to introduce a uniform approach to the selection of families for collaborative studies.¹ It was originally considered that the criteria would be relatively specific rather than sensitive. However, colorectal cancer is a common disease, and the finding of three close relatives affected by bowel cancer within a single large family would not necessarily equate with a specific autosomal dominant disorder, even if one of the subjects happened to be aged below 50 years. Ponz de Leon and colleagues (Gut 2004;53:115–22) show in table 3 that families meeting the Amsterdam criteria but having cancers that are DNA microsatellite stable do not display the clinical features of HNPCC (see below). This is not altogether surprising. In fact it was shown nearly 10 years ago that colorectal cancers in such Amsterdam criteria positive kindreds were frequently DNA microsatellite stable and that the clinical and pathological features of these families were unlike HNPCC (see below).² Of course, the latter findings do not preclude a genetic basis for such clustering of colorectal cancer within a family.

The diagnostic features of HNPCC have accumulated and been refined over time. HNPCC is now defined by a set of clinical, pathological, and molecular features that encompass: a family history of colorectal cancer, a particular spectrum of extracolonic neoplasms, multiple colorectal neoplasia, early onset neoplasia, particular histological features among colorectal cancers, the presence of DNA microsatellite instability, loss of expression of DNA mismatch repair proteins as shown by immunohistochemistry, and a germline mutation in a DNA mismatch repair gene.³ It may not be possible to identify the germline mutation in all families, even when cancers show evidence of deficient DNA mismatch repair. This is merely the result of technical limitations which should not preclude a diagnosis when the other features are met. Close mimicry of HNPCC may occur when hMLH1 has been methylated, perhaps through a familial predisposition to this chemical modification of DNA.⁴ In the latter instance colorectal cancers may develop in a background of multiple hyperplastic polyps.⁵

Now that there is international agreement that the term HNPCC equates to a specific clinicopathological entity,³ there would seem to be little merit in applying the diagnosis when only the limited set of clinical features encompassed by the Amsterdam criteria is met. Reporting of a family history of colorectal cancer has been shown to be unreliable.⁶ What would we think of a judge who chose to base his verdict only on hearsay and rumour and to ignore all evidence of a scientific nature? Why should a family be burdened unnecessarily with the label HNPCC and all its ramifications and how may epidemiological research be advanced by the application of a vague and unreliable diagnostic label?