Purpose: The Surveillance, Epidemiology, and End Results (SEER) program is an important epidemiologic research tool to study cancer. No information is available on its pathologic accuracy for renal cell carcinoma (RCC).
Methods: Central pathology review was analyzed as a part of the United States Kidney Cancer Study. Cases previously identified through the Detroit SEER registry were reviewed. The sensitivity and specificity, and positive and negative predictive values were calculated for each SEER-assigned subtype, with the central review assignments used as the reference.
Results: Of the 498 cases included in this study, 490 (98.5%) were confirmed to be RCC. The overall agreement for histology was 78.2% (κ = 0.55); however, individual cases were frequently reclassified. The sensitivity and specificity for SEER-assigned clear cell RCC were 79.1% and 88.1%, respectively, when based solely on the ICD-O-3 morphology code 8310 (n = 310), and 99.2% and 80.5% when 8312 (RCC not otherwise specified; n = 41) was also assumed to be clear cell. Although RCC not otherwise specified is frequently grouped with clear cell, only 78.1% had this histology. Assignments of papillary and chromophobe RCC had comparable sensitivities (73.5% and 72.4%, respectively) and specificities (97.5% and 97.6%). Positive predictive values for clear cell (excluding/including 8312), papillary, and chromophobe RCC were 95.5%/93.5%, 85.9%, and 65.6%, respectively.
Conclusions: Our findings confirm that nearly all RCC cases are correctly classified in SEER. The positive predictive value was higher for clear cell RCC than for papillary or chromophobe RCC, suggesting that pathologic confirmation may be warranted for studies of non-clear cell tumors.
Keywords: Accuracy; Concordance; Histology; Pathology; RCC; SEER.
Published by Elsevier Inc.