Inherent grading characteristics of individual pathologists contribute to clinically and prognostically relevant interobserver discordance concerning Broders' grading of penile squamous cell carcinomas.

INTRODUCTION

We assessed the reproducibility and prognostic impact of the Broders' grading system (BGS) in a cohort of 147 patients with surgically treated penile squamous cell carcinomas.

MATERIALS AND METHODS

Conventionally stained histology slides were graded according to the BGS in two rounds by two study pathologists. Reproducibility was assessed using ĸ statistics. Multivariable analyses were calculated to predict cancer-specific survival (CSS). The 'mean grade' per pathologist per round was calculated by allocating grade points to each study case (G1-G4: 1-4 points) and dividing the sum of all grade points by the number of cases examined.

RESULTS

The BGS showed substantial interobserver variation (59-87% with ĸ = 0.38-0.69) but almost perfect intraobserver reproducibility (91% with ĸ = 0.86 and 96% with ĸ = 0.94, respectively). The 'mean grade' per pathologist remained nearly constant in both rounds of examination (differences ≤0.05 grade points) but differed between the two pathologists (up to 0.4 grade points). In multivariable analyses, the prognostic impact of the BGS in terms of CSS was strongly pathologist-dependent.

CONCLUSIONS

Clinically and prognostically relevant interobserver discordance concerning the BGS seems, at least in part, to be attributable to inherent 'aggressive' versus 'reserved' grading characteristics of individual pathologists.