Skip navigation

Diagnosing Melanoma

Bandolier was criticised for featuring a US study showing inconsistency between histopathologists expert in the diagnosis of melanoma. It was pointed out that the cases were difficult, and that the inconstancy may not represent the norm for the diagnosis of melanoma. Finding evidence for that conclusion has not been easy, but there is an interesting literature and debate on how to make the histological diagnosis of melanoma better.

CRC Melanoma Pathology Panel

This was a study of 95 sections from pigmented lesions, including equal proportions of benign naevi and primary malignant melanoma, of which half were selected with Breslow thickness <0.76 mm. These were sent to a panel of seven pathologists with a major interest in melanoma plus a dermatopathologist also with a major interest in melanoma.

Slides were sent to panel members before and particularly after a series of meetings to clarify the understanding and use of terms used in making a diagnosis. A series of carefully worded definitions were made for severe nuclear atypia, intraepidermal architectural atypia, an invasive component of radial growth phase, a vertical growth phase or component, mitotic count, regression and established regression.

In a detailed paper with many results the highlights were these:

  • The level of agreement improved after discussion and re-definition of criteria of several features.
  • A high level of agreement was obtained for an overall benign or malignant diagnosis (kappa = 0.77).
  • Use of more specific diagnostic terms with three or four levels of diagnosis resulted in lower levels of agreement.
  • The agreement for individual pathologists reviewing the same slide at different times was high at over 80% for any one, and over 90% on average.

Nationwide survey

One hundred and forty eight UK pathologists participated in two circulations. In the fist circulation of 20 slides no standardised diagnostic criteria were used and the kappa was 0.45 for three categories - benign naevi with no atypia, benign naevi with atypia and melanoma. The same panel as in the CRC study had a kappa of 0.75.

A second survey used 25 slides (19 from the first circulation and six new ones) and used diagnostic criteria of benign, melanocytic intraepidermal neoplasia with or without microinvasion, and melanoma with vertical growth phase. Using these criteria the agreement for pathologists and the panel was the same with a kappa of 0.68.


Both these important papers demonstrate that good agreement can be obtained by the use of standardised criteria. But with melanoma there are always going to be difficult cases where even the experts have problems in making a diagnosis, and where they may disagree.

The main lesson is that discussion and quality assessment is not just for special occasions, but may have to be for ever. The Dutch now have a system where pathologists faced with a difficult case can refer it to one of three experts, who can then refer to two more if the first expert also finds difficulty in making the diagnosis. In a stimulating and thoughtful editorial [3], Mark Cook speaks of the difficulties in continuous quality improvement and in providing the best possible diagnostic service for melanoma. He points out that it will be difficult, and expensive (though how much more expensive than making the wrong diagnosis is not dwelt upon). Just because something is difficult doesn't mean it shouldn't be attempted.


  1. MG Cook, TJ Clarke, S Humphreys et al. The evaluation of diagnostic and prognostic criteria and the terminology of thin cutaneous malignant melanoma by the CRC Melanoma Pathology Panel. Histopathology 1996 28: 497-512.
  2. MG Cook, TJ Clarke, S Humphreys et al. A nationwide survey of observer variation in the diagnosis of thin cutaneous malignant melanoma including the MIN terminology. Journal of Clinical Pathology 1997 50: 202-5.
  3. MG Cook. Diagnostic discord with melanoma. Journal of Pathology 1997 182: 247-9.

previous story in this issue