Reproducibility of histopathologic tumor grading in penile cancer--results of a European project

Since reliable molecular prognostic parameters for inguinal lymph metastases in penile cancer are not available, tumor grading is often used as a surrogate prognostic tool for the indication of inguinal lymphadenctomy and has been integrated into the current TNM classification for penile cancer. The reliability of tumor grading is under discussion. We examined interobserver grading variability in 90 primary penile carcinomas, assessed by 12 different uropathologists from five European countries. Tumor grading, following the CAP scheme, was compared, and interobserver variability was calculated using kappa statistics. The interobserver variability was high as reflected by an overall low kappa coefficient (mean k = 0.34) and reached a moderate level only in 26.4 % of the cases (range 0.02-0.67). The percentage of G1 tumors assigned ranged from 8.6 to 52.5 %, G2 tumors from 27.1 to 72.6 % and G3 tumors from 11.7 to 48.7 %. Only some observers assigned G4 with a range of 0.6-21.9 %. Subdivision into low and high grade according to UICC and EAU classifications differed significantly (P < 0.001). Low reproducibility of grading in penile carcinomas with the favored method does not allow a reliable prognostication of tumor aggressiveness. Inclusion of histological grading into the TNM classification currently seems not to be a benefit.