Award Number : DAMD 17 - 00 - 1 - 0029 TITLE : Prostate Cancer in Nigerians , Jamaicans and U . S . Blacks PRINCIPAL INVESTIGATOR :

Our aim was to study the level of interobserverconcordance in the Gleason scores of prostate needlebiopsy specimens reported at 1 institution. Aretrospective review of all prostate needle biopsyspecimens in which a diagnosis of adenocarcinoma wasmade during the year 2000 was conducted. Parametersevaluated included the Gleason score, Gleason gradesidentified, the percentage of Gleason grades 4 and 5,and the percentage of tumor in the biopsy specimen.Our results demonstrated a 60% overall concordance inconsensus Gleason scores, which increased to 80%when considered in groups of a Gleason score of lessthan 7 vs 7 or more. The greatest discordance seemedto be in distinguishing Gleason score 6 from 7 and wasmore frequent among biopsy specimens with lowertumor volumes, particularly among those with less than30% involvement. A small percentage of Gleason grade4 pattern might predict disagreement as well. Strategiesfor improving accuracy of Gleason score 7 should bedevised, and consensus diagnosis for biopsy specimensthat demonstrate a low percentage of tumor volume isrecommended.The Gleason histologic grade of prostate adenocarci-noma is thought to be one of the most powerful predictors ofbiologic behavior and often has an important role in deter-mining patient treatment. It is well known among patholo-gists that assessment of this histologic grade is associatedwith interobserver variation. However, the usefulness of anygrading system must depend on reasonable interobserverconcordance. With this in mind, we sought to study the levelof interobserver agreement in the Gleason scores assigned toprostate needle biopsy specimens and reported at the Univer-sity Hospital of the West Indies, Kingston, Jamaica. Materials and Methods We prospectively obtained all prostate needle biopsyspecimens for which a diagnosis of carcinoma was made atthe University Hospital of the West Indies during the year2000. This diagnosis was made by any of 9 staff anatomicpathologists (including one of us [K.C.C.]), who received thebiopsy specimens while on surgical pathology sign-out dutiesaccording to the service roster. One of us (K.C.C.) thenreviewed all biopsy specimens without knowledge of theprevious Gleason score. This pathologist, by virtue of partici-pation in an international collaborative project on prostatecancer, has had the added advantage of reviewing a largenumber of such biopsy specimens. Moreover, the interob-server variation for this pathologist previously was comparedwith that of pathologists from 3 different countries, with agood level of concordance (κ = 0.3317; P = .0173; and κ =0.2437; P < .0001, respectively, for individual Gleason sumsand various sum groupings).1 Am J Clin Pathol 2004;122:373-376 373373 DOI: 10.1309/MHCY35FJ296CLLC8 373© American Society for Clinical Pathology Coard and Freeman / GLEASON GRADING REPRODUCIBILITY During the histopathologic review, the following factorswere evaluated: Gleason score, Gleason grades, percentageof Gleason 4 and 5 pattern, and overall percentage of tumorin the biopsy specimen. Evaluation of concordance of thebiopsy score with that of a subsequent radical prostatectomyspecimen was not undertaken. Statistical Analysis Means and SEs were calculated for selectedhistopathologic characteristics. Because dichotomizingGleason scores into those less than 7 and those 7 or morehelps distinguish between cancers with relatively favorableand those with unfavorable prognoses, respectively,2disagreement in the assignment of Gleason scores of 6 and7 can be clinically significant. Therefore, interobserveragreement across these categories was assessed by usingthe weighted κ statistic, and histopathologic correlates ofdisagreement were determined by using the 2-sample t testand logistic regression analysis.