Reproducibility of pathologic scoring systems for periprosthetic adverse local tissue reactions: A cross-sectional study

Abstract

BackgroundSeveral previous studies have described broad histologic classifications of peri-prosthetic reactions that likely reflect the underlying mechanism of arthroplasty failure; however, a consensus has not yet been reached about the relative importance of individual observations. Question/PurposeThe purpose of this study was to evaluate the inter-examiner repeatability of commonly used histopathologic grading methods, and to determine the utility of assigning a more simple, global categorization in patients undergoing revision THA surgery of implants with a variety of bearing combinations. MethodsBetween March 2013 and February 2020, a total of 2131 patients underwent revision hip arthroplasty surgery at a one center, of which 12% (248 of 2131) of patients were enrolled. Two pathologists independently reviewed microscope slides of periprosthetic tissue from these patients, of which 425 slides (229 hips, 222 subjects) were reviewed by both pathologists. Separate slides were used for a priori training of the pathologists. Slides were evaluated with the Campbell Aseptic Lymphocyte-dominant Vasculitis-Associated Lesion (ALVAL) score, the Oxford ALVAL score as modified by Grammatopolous, the Fujishiro and Natu scores, and a proposed simplified pattern classification, similar to that of Krenn et al., that incorporates individual factors of these existing scoring methods and was previously shown to correspond to Magnetic Resonance Imaging findings. Inter-rater agreement was assessed using Gwet’s AC1 and AC2 coefficients and correspondence analysis was used to examine associations between individual factors of prior scoring methods with the proposed major pattern. ResultsAlmost perfect inter-rater repeatability (Gwet’s AC2 > 0.8) was found for 71% (15/21) of the individual factors, and substantial interrater agreement was found for the proposed major overall pattern (Gwet’s AC1: 0.80, 95%CI: 0.72–0.85). Correspondence analysis was able to explain 89–91% of data variability and was able to identify individual features not commonly associated with a major pattern, but discriminatory of the major pattern, such as “Lymph Cuff Thickness 0.25–0.5″ with ALVAL. ConclusionIn contrast to prior examinations, excellent interrater agreement was found that may be attributable to a priori training of raters with a test set of slides or difficulty of interpreting grading criteria. The proposed simplified major pattern classification may facilitate evaluation of histopathologic tissue samples.