Background/Purpose: Musculoskeletal (MSK) manifestations are common in systemic lupus erythematosus (SLE) and other ANA-associated rheumatic musculoskeletal diseases (ANA-RMDs). Presently clinical trials recruit patients from individual diagnoses. Recruitment is often difficult, effect sizes are small and licensed therapies are only available to patients with the most common ANA-RMD, SLE.

Objectives:
1. To compare the symptom impact and immunophenotype of arthritis associated with a range of ANA-associated autoimmune diseases
2. To explore alternative classifications of the spectrum of ANA-associated arthritis to define basket populations for trials
3. To validate the MSK-BILAG across multiple diagnoses

Methods: We used analysis of flow cytometry for major subsets, gene expression (signatures for interferon, plasma blast, neutrophil, myeloid inflammation and erythropoiesis) and patient reported data from a mixed cohort of ANA-associated autoimmune diseases. 215 patients with MSK symptoms were included, 90 with SLE and 125 with an another diagnosis (UCTD, pSS, IM, MCTD and SSc).

The implications of legacy diagnosis membership were compared using QoL measures (SF36, EQ5D-5L, Patient VAS, ICECAP-A, FACIT-F, WPAI and ICECAP-A. Gaussian mixture modelling was used to explore alternative classifications more suitable for basket trials. The association of BILAG-MSK with physician VAS was tested in SLE and the non-SLE group.

Results: Legacy diagnosis was a poor indicator of the patient experience with Kruskal Wallis testing revealing only 2 significant differences (EQ5D Mobility and EQ5D-5l index domains) between diagnoses. Flow cytometric and gene expression data showed no significant differences between legacy diagnoses.

Better predictors of the patient experience were chromatin antibody positivity and non-European ancestry which were associated with worse EQ5D-5l index scores (P < 0.05 in both).

Preliminary gaussian mixture modelling analysis of this population based on variables identified as important within the principle component analysis identified 4 clusters. A biologically active MSK symptom dominant cluster (n=140), a biologically active non MSK symptom dominant cluster (n=43), a biologically and symptom quiescent cluster (n= 24) and a biologically and symptom quiescent cluster (n=65).

The application of the BILAG-MSK domain across ANA-RMDs correlated well with the Physician General Assessment scores of MSK disease activity suggesting it may be reasonable to apply this measure in any future basket trial.

Conclusion: Inflammatory arthritis is similar in both clinical impact and immune phenotype across ANA-associated RMDs, suggesting similar benefits of therapy for this feature within each disease. Machine learning reveals an alternative basket of patients suitable for clinical trials who have clinical synovitis, high symptom burden and consistently increased markers of immune activation, and a mixture of legacy diagnoses, who are more numerous than MSK SLE alone. A further smaller basket of patients have high symptom burden but little evidence of active inflammation.

ACR Meeting Abstracts - https://acrabstracts.org/abstract/defining-a-basket-population-for-ana-arthritis-trials/