intraepithelial neoplasia

Downloaded from jcp.bmj.com on July 15, 2011 - Published by group.bmj.com

GClin Pathol 1995;48:59-60

59

Pathologist variation in reporting cervical borderline epithelial abnormalities and cervical intraepithelial neoplasia T Creagh, J E Bridger, E Kupek, D E Fish, E Martin-Bates, M J Wilkins

Abstract Aims-To evaluate the interobserver variation in the diagnosis of cervical intraepithelial lesions, including the new category "borderline abnormalities of uncertain significance" (BAUS) which has not been tested before. Methods-Biopsy specimens of 122 patients were reviewed by five histopathologists and the diagnoses subjected to K statistical analysis. Results-There was poor interobserver agreement in all categories, particularly between BAUS and normal tissue. Conclusions-The current guidelines for the histological diagnosis of cervical intraepithelial neoplasia and BAUS are poorly reproducible. (J

Clin Pathol 1995;48:59-60)

Keywords: Cervical intraepithelial neoplasia, variation, diagnosis.

Department of Histopathology and Cytopathology, St Mary's Hospital, Praed Street, London W2 1NY T Creagh M J Wilkins E Martin-Bates Department of

Histopathology, The Royal Hospital, Wolverhampton J E Bridger Academic Department of Public Health, St Mary's Hospital Medical School, Norfolk Place, London E Kupek

Department of Histopathology, East Surrey Hospital, Redhill, Surrey D E Fish Correspondence

to:

Dr T Creagh. Accepted for publication 1 July 1994

At present, there is considerable disagreement concerning the diagnosis and grading of cervical intraepithelial neoplasia (CIN), with some observers favouring a two grade and others a four grade system. The interobserver variation in the diagnosis of low grade dysplasia (CIN I) and high grade dysplasia (CIN II and CIN III) has previously been shown to be considerable by two independent groups,' 2 with poor agreement in the diagnosis of CIN I and CIN II and mediocre agreement for CIN III. Despite these results it has been proposed recently that the standard classification of CIN should not only be retained but that a new category, "basal abnormalities of uncertain significance" (BAUS), should be introduced.3 The purpose of this study was to evaluate interobserver variation for this proposed classification.

Methods The study population comprised 122 consecutive patients who had had cervical biopsy specimens taken at colposcopy. The specimens were received fixed in 10% formaldehyde solution and were processed routinely. The same set of slides was reviewed by five histopathologists and each allocated to one of six diagnostic categories: normal squamous epithelium, BAUS, and CIN I, II and III, together with a separate category for human papilloma virus (HPV) without CIN. The diagnostic criteria used were those defined by Buckley et al4 for CIN and by Anderson et al3 for basal cell abnormalities and HPV changes alone. The degree of agreement between random pairs of observers was evaluated using the K statistic.5 This statistic is an index of interobserver agreement which has been corrected for chance and thus gives the degree of interobserver agreement over and above that which would be expected by chance alone. Values greater than 0 75 represent excellent agreement, between 0-4 and 0 75 fair to good agreement, and below 0O4 poor agreement beyond chance. Kappa values were calculated for each diagnostic category and for overall agreement. A weighted K value was also calculated for overall agreement to take the more serious disagreement between widely differing, as opposed to neighbouring, categories into account.6

Results Five observers examining 122 cases gave a total of 610 results which are presented in the table. To characterise the degree of agreement between pairs of observers, a symmetrical agreement matrix was formed which had a total of 2440 possible pairs (122 by five by

Agreement between five observers in categorising 122 histological specimens HPV BA US N Normal CIN I Diagnosis Normal HPV Borderline CIN I CIN II CIN III Total

20 133 83 186 102 86 610

2

41 19 18 0 0 80

N = number of individual observations.

41 254 107 115 10 5 532

19 107 62 125 19 0 332

18 115 125 332 117 37 744

CIN II

CIN HI

Total

K value

0 10 19 117 162 100 408

0 0 0 37 100 202 244

80 532 332 744 408 344 2440

0 009 0-333 0.060 0-203 0-276 0-519 0-232


Creagh, Bridger, Kupek, Fish, Martin-Bates, Wilkins

60

four). In the table-for example, 86 individual observations were categorised as CIN III, each of which may be compared with each of the four diagnoses made by the other observers on the same specimen, resulting in a total of 344 (86 by four) comparisons. In 202 instances the other observer made a diagnosis of CIN III, in 100 instances CIN II, in 37 instances CIN I, and in five instances HPV. The final column in the table gives K values for each category. The degree of agreement was relatively good for CIN III, poor for HPV, CIN I and II, and very poor for normal and borderline cases. A weighted K value (0 504) for the whole range of categories showed moderate agreement. Discussion The results show poor interobserver agreement in the histological diagnosis of HPV, CIN I, and CIN II but relatively good agreement for CIN III. The most significant finding was the very poor agreement in the diagnosis of both the recently proposed category, BAUS, and normal epithelium. As shown in the table, there was difficulty in distinguishing both normal epithelium and BAUS from HPV and CIN I. These findings are not altogether surprising in view of the fact that two previous studies have shown poor interobserver agreement in the histological diagnosis of CIN. In one study' this was particularly marked in the diagnosis of CIN I and reactive proliferations; in the other,2 there was difficulty in distinguishing between CIN I and CIN II and the presence of HPV infection. Instead of clarifying the situation the introduction of the category BAUS has made it even more difficult for individual pathologists to accurately categorise these lesions. Possible explanations for the results are, firstly, although the morphological criteria for the diagnosis of CIN appear to be well defined, in practice there are widely differing

interpretations by individual pathologists. Secondly, CIN represents a continuous spectrum of increasing abnormality with differing grades present in any one specimen. Thirdly, HPV related changes may also be present and may make the application of any diagnostic criteria even more subjective. This final point is illustrated by the fact that in five instances any one observer made a diagnosis of HPV alone when the others had diagnosed CIN III. This is reflected in the moderate overall weighted K score. In each of these cases there was particularly florid HPV infection, which is well recognised as a cause of diagnostic difficulty.7 A pathological classification should be relevant, easily understood, highly reproducible, and clinically useful. Our study has confirmed that not only are the current criteria for the histological diagnosis of CIN poorly reproducible but that the proposed new classification, with the addition of BAUS as a separate category, is even less so. Whether the classification will prove clinically useful remains to be seen; however, it must be borne in mind that a further abnormal diagnostic label which can be applied to a patient is only likely to increase psychological stress and also increase treatment of transient minor abnormalities.

1 Ismail SM, Colclough AB, Dinnen JS, Eakins D, Evans DMD, Gradwell E, et al. Observer variation in histopathological diagnosis and grading of cervical intraepithelial neoplasia. BMJ 1989;298:707-10. 2 Robertson AJ, Anderson JM, Swanson-Beck J, Burnett RA, Howatson SR, Lee FD, et al. Observer variability in histopathological reporting of cervical biopsy specimens. J Clin Pathol 1989;42:231-38. 3 Anderson MC, Brown CL, Buckley CH, Fox H, Jenkins D, Lowe DG, et al. Current views on cervical intraepithelial neoplasia. J Clin Pathol 199 1;44:969-78. 4 Buckley CH, Butler EB, Fox H. Cervical intraepithelial neoplasia. Y Clin Pathol 1982;35:1-13. 5 Fleiss JL. Statistical Methods for Rates and Proportions. 2nd edn. New York: Wiley, 1981:212-34. 6 Fleiss JL, Cohen J. The equivalence of weighted kappa and intraclass correlation coefficient as measures of reliabil-

ity. Educational Psychol Measurements 1973;33:613-19. 7 Schneider V. Microscopic diagnosis of HPV infection. Clin Obstet Gynaecol 1989;32:1358-65.


Pathologist variation in reporting cervical borderline epithelial abnormalities and cervical intraepithelial neoplasia. T Creagh, J E Bridger, E Kupek, et al. J Clin Pathol 1995 48: 59-60

doi: 10.1136/jcp.48.1.59

Updated information and services can be found at: http://jcp.bmj.com/content/48/1/59

These include:

References

Article cited in: http://jcp.bmj.com/content/48/1/59#related-urls

Email alerting service

Receive free email alerts when new articles cite this article. Sign up in the box at the top right corner of the online article.

Notes

To request permissions go to: http://group.bmj.com/group/rights-licensing/permissions

To order reprints go to: http://journals.bmj.com/cgi/reprintform

To subscribe to BMJ go to: http://group.bmj.com/subscribe/