FRI0505 CORRELATIONS BETWEEN CARTILAGE MOLECULAR COMPOSITION DETERMINED BY RAMAN SPECTROSCOPY AND MANKIN SCORE: IMPACT OF INTER AND INTRA-VARIABILITY

Abstract

Background Osteoarthritis (OA) is an incurable disease and methods for its early diagnosis are still an unmet clinical need. Raman spectroscopy (RS) presents potential as a diagnosis technique based on the detection of peaks that can be assigned to cartilage components and molecular rearrengements during disease progression.1Mankin score (MS) is the main validated method to evaluate severity of cartilage degradation considering structure, cell distribution, Safranin-O staining and tidemark integrity.2 Objectives To evaluate the correlations between OA cartilage RS assigned peaks and MS, considering the inter- and intra-variability of different observers. Methods MS analysis (Subscore-I, structure: 0-6; -II, cellularity: 0-3; and -III, safranin-O staining: 0-4; Total Score: 0-13) of human OA cartilage explants from 22 donors (age range: 32-92), obtained from lesion (n=22) and/or adjacent tissues (n=14), was performed by 3 blinded observers (O1, 2 and 3). Moreover, one of the observers performed the scoring in triplicate, with at least one month between observations. Inter- and intra-observer variability was determined by kappa and intraclass correlation (ICC) coeficients. Raman spectra were obtained with a FT-Raman Bruker RFS100 (λ=1064nm) and main peaks assigned (6 ratios related to proteoglycans, collagen, lipid index or calcium phosphate). Spearman’s nonparametric correlation coefficient rho was used to compare MS and RS assigned peaks. Results Inter-observers variability indicated good (ICC>0.74) or moderate agreement (ICC>0.5) for all scores in lesions, whilst only a good agreeement (ICC=0.70) was found for subscore-I, in adjacent tissues, and no agreeement for the remaining parameters (subscores -II, -III, and total scoring). However, when performing analysis using kappa coefficients, a simultaneous agreement between the 3 observers was not observed. Intra-observer variability revealed good concordance (ICC>0.6) for all subscores and total scoring in cartilage for both sites, except for subscore –III, in adjacent tissues. In this case, ICC results were confirmed by kappa coefficients. Spearman’s correlation coefficient between cartilage main peaks assigned by RS and MS indicated significant differences between observers(Fig.1). Correlations were found for a greater number of MS subscores in O1 (6) regarding O2 (4) or O3 (3) which could be related to the observers’ experience (being O1>O2>O3). These correlations were mostly found in lesions (5, 3 and 1 for O1, 2 and 3, respectively) in comparison to adjacent tissues (2, 1 and 2 for O1, 2 and 3, respectively). Conclusion Eventhough inter-observation correlations for MS were in the moderate- good range, when analyzing kappa coefficients (categorical variables) these were not maintained. In addition, inter- and intra-observer variability results for adjacent tissues revealed possible limitations when characterizing early to mild OA. In view of MS-RS correlations, a reader dependency is underlined, indicating MS subjectivity and further limitations in the validation of RS using MS.