Background/Purpose: Acro-osteolysis (A-O, terminal tuft digit resorption) is characteristic of systemic sclerosis (SSc), occurring in 20-25% of patients (1). Outcome measures are much needed. Following our previous work (2) that developed a scoring system for A-O, our objective was to examine different methods of scoring A-O over time, specifically comparing subjective clinician opinion to more formal digit-by-digit scoring, to investigate their potential as outcome measures.

Methods: Patients identified from a tertiary referral centre for SSc who had a first visit between 2006-2018, who had suitable radiographs (both hands, two time points). The final image set contained 180 image pairs from 90 patients. Radiographs were presented to a single observer for assessment:1. 180 single (blinded) images; 10 fingers/digits scored: 0 (normal) to 4 (complete terminal tuft resorption); total (sum of all digits), maximum, and Johnstone scores (2) calculated.2. 90 image pairs presented (left vs right) in a randomised order. Images reviewed answering “is the acro-osteolysis status of the right-hand image more or less severe than the left-hand image?”: 5-point ordinal scale: much less severe (-2) to much more severe (+2). Images with fewer than 3 classifiable digits per hand were excluded. Missing values in included images were replaced with the average score (mean of all other digits). For paired images the clinician change scores were used to dichotomise: “changers” (patients assessed as changing over time) who were scored “much more severe” (+2) or “more severe” (+1) and “no change” scoring “same” (0), “less severe” (-1), or much less severe (-2). We derived “change over time” data between pairs of images – with change calculated in (1) total score, (2) maximum score, and (3) Johnstone score. We investigated differences in scoring between the “changers” vs. “no change” groups using T-tests (total) or Mann-Whitney tests (maximum, Johnstone).

Results: Patient demographics (n=90) are presented in Table 1. No score could be given in 102/1800 digits assessed due to amputation (n=17, 17%) or obscured digit (n=85, 83%). 10/180 images, and 9/90 pairs were excluded due to missing data. 81 image pairs were analysed after meeting our criteria, with 25 ultimately labelled as “changers”, and 56 labelled as “no change”. Figure 1: box plots comparing change-over-time data (total, max, or Johnstone) between the two groups from the dichotomised subjective clinician change score. A significant difference in change in total score of >6 points is identified between the two groups.

Conclusion: Total score appears to show promise for measuring clinically relevant change in A-O. Maximum and Johnstone scores are less sensitive with no obvious difference in group medians (despite statistical significance), likely due to the relatively large number of cases with zero change in both groups. We have explored pragmatic options for quantifying clinically relevant change in A-O with relevance to future longitudinal studies including impact of treatment. References1. Avouac, et al. Ann Rheum Dis 2006:1088–92. 2. Johnstone, et al. Rheumatology 2012;:2234–8.

Table 1. Patient demographics.

Figure 1. Boxplots comparing changers and no change groups as defined by clinician opinion (paired image assessment): a) Change in total score, b) change in maximum score, and c) change in Johnstone score. Comparisons of means/medians for a, b and c are shown bottom right, with p-values from T-test or Mann-Whitney tests as appropriate.

« Back to ACR Convergence 2025

ACR Meeting Abstracts - https://acrabstracts.org/abstract/investigating-scoring-systems-to-measure-clinically-relevant-changes-in-systemic-sclerosis-related-finger-acro-osteolysis-over-time-as-assessed-in-serial-hand-radiographs/