交界性导管增生性乳腺病的病理诊断重复性研究

目的探讨病理医师之间对交界性导管增生性乳腺病的诊断重复性及其影响因素，寻求提高诊断重复性和准确性的措施。方法参照Page标准收集43例交界性导管增生性乳腺病病例，每例选取一张切片并随机排序。十位非乳腺专科病理医师独立盲法读片后从以下六种诊断中选取一种：轻度普通型增生、中-重度普通型增生、轻度非典型增生、中-重度非典型增生、导管原位癌和导管原位癌伴浸润。并将观察者按从事病理诊断的年限均分为高年资组和中低年资组。对十位病理医师和不同年资组间的诊断重复性进行Kappa分析。同时将两位乳腺专科病理医师按照Page标准共同确认的诊断作为参照，对十位病理医师的诊断准确性和过度诊断进行统计学分析。结果十位病理医师间对交界性导管增生性乳腺病的诊断重复性较低，特别是非典型增生。高年资医师组的诊断重复性和准确性略高于中低年资医师组。部分医师存在不同程度的过高或过低诊断。随着诊断类别的简化，诊断重复性会提高。结论统一运用客观准确的诊断标准是提高诊断重复性和准确性的重要措施。

Interobserver Reproducibility in the Pathologic Diagnosis of Borderline Ductal Proliferative Breast Diseases

Objective The aim of this study involving general pathologists was to assess interobserver reproducibility in the pathologic diagnosis of borderline ductal proliferative breast diseases. Methods Ten general pathologists independently reviewed 43 specimens chosen to represent the spectrum of borderline ductal proliferative breast lesions. All slides were blindly reviewed without given standardized criteria, and were classified as either mild usual hyperplasia, moderate-severe usual hyperplasia, mild atypical hyperplasia, moderate-severe atypical hyperplasia, ductal carcinoma in situ, or ductal carcinoma in situ with invasion. According to the years of training, these 10 general pathologists were divided into the experienced group and the less trained group. Interobserver agreement was statistically analyzed using Kappa statistic. Then, by comparing all the diagnoses of individual pathologist with the consensus opinion confirmed by two breast pathologists in terms of Page standard, we acquired the diagnostic accuracy and ascertained the undue diagnosis. Results The ten general pathologists' interobserver reproducibility in the diagnosis of borderline ductal proliferative breast diseases was rather low, especially that in their diagnosis of atypical hyperplasia. The reproducibility and accuracy were slightly higher in the experienced pathologists than in the less trained pathologists. Some pathologists made over-diagnosis or under-diagnosis of the lesions to different degrees. However, when the categories of diagnositc terms were simplified, the interobserver reproducibility increased. Conclusion The use of standardized criteria is an important approach to increasing the diagnostic reproducibility and accuracy.