Steinberg D M, Sauvageot J, Piantadosi S, Epstein J I
Department of Pathology, Johns Hopkins Hospital Medical Institutions, Baltimore, Maryland 21287, USA.
Am J Surg Pathol. 1997 May;21(5):566-76. doi: 10.1097/00000478-199705000-00010.
Prior studies have not analyzed grading patterns and accuracy in nonacademic sites and have not analyzed reasons for discrepant grades. We analyzed 499 radical prostatectomy (RP) specimens at The Johns Hopkins Hospital (JHH) from 1994 and compared them with the corresponding JHH needle biopsy Gleason grade and, when available (n = 390), to the outside institution (non-JHH) Gleason biopsy grade. For JHH, there was exact agreement between biopsy and RP in 58% and agreement to within one digit in 93% of cases, compared with 34% and 67%, respectively, for non-JHH. Combining cases into more meaningful groups (Gleason 2-4, 5-6, 7, and 8-10), there was 66% exact correlation between the biopsy and RP for JHH as compared with 45% for non-JHH. Non-JHH undergraded biopsy results more than JHH, with 22.3% and 1.2% Gleason score 2-4, respectively. None of the cases with a Gleason score of 2-4 on biopsy from JHH or non-JHH were Gleason score 2-4 on RP. Fifty-five percent of the tumors with non-JHH needle biopsy results graded Gleason 2-4 had either capsular penetration, seminal vesicle, or lymph node involvement. All of the tumors with needle Gleason score 2-4 at JHH were organ confined. The JHH needle biopsy grade correlated better with pathologic stage than did non-JHH (R = 0.27 JHH vs. R = 0.12 non-JHH). Extent of cancer in the biopsy sample was not a factor in the accuracy of predicting RP grade or stage. Eighty-two cases evaluated at JHH were signed out by a genitourinary pathologist; the grading of the biopsy samples by other JHH pathologists was just as accurate. Gleason score of > or = 7 on the biopsy sample predicted a Gleason score of > or = 7 in the RP 87.5% of the time. A Gleason score of < 7 predicted a Gleason score of < 7 only 63.9% of the time. Discordant grades in some cases reflected patterns of cancer on needle biopsy that were borderline between two different Gleason scores. Sampling was the major source of discrepancy and was often due to the high-grade component in the RP not being present in the biopsy results or due to a component of cancer on the needle reflecting such a small percentage of the pattern seen on RP that this pattern was not included in the final RP Gleason score.
既往研究未分析非学术机构的分级模式及准确性，也未分析分级不一致的原因。我们分析了1994年以来约翰·霍普金斯医院（JHH）的499例根治性前列腺切除术（RP）标本，并将其与相应的JHH穿刺活检Gleason分级进行比较，如有外部机构（非JHH）的Gleason活检分级（n = 390），也一并进行比较。对于JHH，活检与RP分级完全一致的病例占58%，相差不超过一级的病例占93%；相比之下，非JHH的这两个比例分别为34%和67%。将病例合并为更有意义的组（Gleason 2 - 4、5 - 6、7以及8 - 10）后，JHH活检与RP分级的完全相关性为66%，而非JHH为45%。非JHH对活检结果的分级低于JHH，非JHH和JHH的Gleason评分2 - 4分别为22.3%和1.2%。JHH或非JHH活检Gleason评分为2 - 4的病例，RP时均未出现Gleason评分2 - 4。非JHH穿刺活检Gleason分级为2 - 4的肿瘤中，55%有包膜侵犯、精囊或淋巴结受累。JHH穿刺Gleason评分为2 - 4的所有肿瘤均局限于器官内。JHH穿刺活检分级与病理分期的相关性优于非JHH（JHH的R = 0.27，非JHH的R = 0.12）。活检样本中的癌症范围不是预测RP分级或分期准确性的因素。JHH评估的82例病例由泌尿生殖病理学家给出诊断结果；其他JHH病理学家对活检样本的分级同样准确。活检样本Gleason评分≥7时，RP时Gleason评分≥7的预测准确率为87.5%。活检样本Gleason评分<7时，RP时Gleason评分<7的预测准确率仅为63.9%。部分病例分级不一致反映了穿刺活检的癌症模式处于两种不同Gleason评分的临界状态。取材是分级不一致的主要原因，通常是由于RP中的高级别成分未出现在活检结果中，或者穿刺活检中的癌症成分在RP所见模式中占比过小，以至于该模式未纳入最终的RP Gleason评分。
