Interobserver agreement of scoring of histopathological characteristics and classification of lupus nephritis.

BACKGROUND Assessing renal biopsies from patients with lupus nephritis (LN) is a difficult task and it is subject to interobserver variability. In this study the interobserver agreement amongst five nephropathologists was analysed. METHODS Five specialized nephropathologists scored 126 biopsies, comprising 87 first and 39 repeat biopsies from 87 patients with biopsy-proven proliferative LN, included in a randomized controlled trial. The interobserver agreement [expressed as intraclass correlation coefficients (ICC)] of the scored histopathological items was calculated. Also, the WHO1995 and ISN/RPS2003 classification systems for LN were compared, with extra attention being given to the comparison between patients with diffuse proliferative LN with either segmental (IV-S) or global (IV-G) lesions. RESULTS There was a wide range of agreement. A good interobserver agreement (ICC>0.6) was present in 15%, and a moderate interobserver agreement (ICC 0.4-0.6) in 31% of the scored items. The activity index for LN showed a good (ICC 0.716) and the chronicity index a moderate (ICC 0.494) interobserver agreement. Both classification systems showed low agreement, although consensus was easily reached. Patients classified as IV-S (n=15) had more favorable clinical parameters at study entry than those with class IV-G (n=57). Although suggested by others, we found no differences in outcome between these two subclasses. CONCLUSIONS This study shows that, although definitions were agreed upon beforehand, even specialized on nephropathologists have difficulties with scoring histopathological characteristics of LN, particularly with SLE the classification systems.