Prognostic factors in invasive breast carcinoma The management of patients with breast carcinoma has changed radically in the last decade for a number of reasons. The range of therapeutic options for both local and systemic therapy has widened considerably with choices between conservation surgery (with or without post-operative irradiation) and mastectomy and hormonal or cytotoxic therapy. It has also been shown unequivocally that invasive breast carcinoma is not a single entity, but should be regarded as a neoplastic spectrum with a prognostic range from excellent to poor. 43, 44 Improved health education and the introduction of mammographic screening have led to the identification of a greater number of cancers at the better end of this prognostic range and it is self-evident that such patients require different treatment to those with more aggressive disease. It is therefore increasingly important that clinicians are given the most accurate prognostic information on which to base the appropriate therapy for women on an individual basis. In this way both over and under treatment can be avoided. This clinical need for information has fostered an ever widening search for the ideal prognostic factor, but despite an extensive literature devoted to newer techniques such as molecular biology, traditional histopathological examination of breast cancer specimens still yields the most useful information. It is now recognised throughout the world that the minimum data set provided by histopathologists should comprise an accurate measurement of tumour size, histological type and grade, assessment of lymph node status and a record of the presence or absence of lymphovascular invasion. 6, 7, 45 Although there are no specific diagnostic pitfalls in the assessment of these factors studies designed to examine the accuracy of histopathological reporting, particularly those associated with the UK NHSBSP and the European Group for Breast Screening Pathology, have highlighted problems in consistency and reproducibility. 1-3 In order to assess the accuracy of assessment of tumour size the participants in the UK scheme were asked to measure the maximum extent of the invasive component on the histological sections. 1 It was assumed, somewhat naively, that this would be the most consistent of all the factors, but, initially, large variations were encountered. Although 80% of measurements were within + 3 mm of the median value some results were well outside this range, eg, estimates of 5 and 22 mm recorded for a 15 mm tumour. More detailed instructions were then issued, 7 with explanatory diagrams and these have lead to improvements in this area (data as yet unreported). The level of consistency in reporting the morphological subtypes was disappointingly poor in the early reports from the UK scheme, with an overall kappa value of 0.21, a result which could almost have been achieved by chance alone. Similarly, poor consistency for histological grading was obtained by the non-co-ordinators, although the co-ordinators, with more experience in breast pathology, achieved a value of 0.46, in the average range. The results for tumour type were, perhaps, understandable, because at the time of the original circulations, now over eight years ago, the scheme had not focused specifically on this issue and few pathologists recorded this feature routinely. The poor performance with grading was rather more of a concern, since the method recommended is summarised in the document Pathology Reporting in Breast Cancer Screening, 7 now in its second edition and had been shown to have an acceptable level of consistency and reproducibility in several other studies. 46-48 It is likely that, as with type, relatively few pathologists were grading breast cancer at the time of the study and it is also possible that too little attention has been paid by participants to the published criteria. Better levels of agreement have been achieved in the European scheme, involving only 20-25 pathologists, all with a specific interest in breast pathology. 3 There is also evidence from the UK scheme that participation itself improves performance. Indeed now that both schemes have been in operation for a number of years it appears from recent unpublished results that kappa values for both typing and grading of about 0.6, in the 'good agreement' range, can be obtained. These results are extremely encouraging and there seems to be little doubt that consistent and reproducible results can be obtained for the traditional pathological prognostic factors, especially if agreed diagnostic protocols, as recommended in the European Community guidelines are used. It is also clear that attention now needs to be paid to accuracy of the pathological assessment of lymph node status, especially in view of the development of sentinel node biopsy. 49 The management of individual patients is increasingly dependent on these factors and pathologists have a duty of care to ensure that they provide the most accurate information possible. |
Copyright 2000, The Author(s) and/or The Publisher(s)
| Organisation: FORPATH asbl  |
Coordination: Dr Bernard Van den Heule |
Host: Labo CMP |