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

Abstract

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.