Core biopsies of renal tumors: a study on diagnostic accuracy, interobserver, and intraobserver variability

The diagnostic accuracy of in-bench core biopsies (CBs) from renal masses, and the interobserver and intraobserver variability in pathological subtyping of renal tumors were assessed.

We performed two CBs in 62 consecutive renal masses suspected for renal cell carcinoma (RCC), obtained after radical or partial nephrectomy and, in one case, after autopsy. Routine hematoxylin-eosin (HE)-stained sections from each CB were evaluated by five pathologists on two occasions. The surgical specimen was the reference standard. Diagnostic accuracy and the generalized kappa for intraobserver and interobserver agreement were calculated.

Five tumors were benign and 57 malignant. Eight percent to 16% of the CBs were considered inadequate for diagnosis. In 0-8% of the cases, the pathologist could not discriminate between a benign or malignant tumor. Overall accuracy ranged from 77% to 90%. Sensitivity (79-100%) and positive predictive value (100%) were high with narrow 95% confidence interval (95%CI). Specificity (100%) was high but negative predictive value (29%-100%) varied, with wide 95% CI. Interobserver agreement was fair to almost perfect (kappa=-0.010 to 0.830) for the different subtypes. In 64-81% of the CBs, the subtype was correctly classified. Intraobserver agreement was substantial (mean kappa=0.628) for all pathologists.

Diagnostic accuracy of CBs was high, with a diagnostic yield varying between 84% and 92%. Pathological subtyping of CBs was highly reproducible with the exception of the chromophobe renal cell carcinoma, which was problematic on HE-stained sections only.