Innovative research approaches bring us closer to precision medicine for inflammatory rheumatic diseases

As we look towards the future of immunology research with an eye for innovation and improvement, we are continuously examining how we create the most practical and informed approaches to assessing the health of those with immune-mediated inflammatory diseases and their treatment response patterns. During the EULAR Congress, where Janssen is presenting 38 abstracts, I wanted to provide some context to the research we’re pursuing to more precisely target immune-mediated diseases like psoriatic arthritis (PsA), using novel composite endpoints and machine learning.

The value of composite endpoints in rheumatology

The complexity of the pathogenic processes underlying inflammatory rheumatic diseases makes it difficult to use a single outcome measure to understand a patient’s response to treatment. These diseases can be multifaceted, with patients presenting with varied symptoms. Because of this, they often require measurement of multiple outcomes (1).

PsA is a chronic inflammatory disease with multiple manifestations, including peripheral joint inflammation, enthesitis (pain where the bone, tendon and ligament meet), dactylitis (severe inflammation of the fingers and toes), axial disease, and skin lesions associated with plaque psoriasis (2-5). The type and severity of skin and joint symptoms is highly variable, and there is no correlation between these symptoms (3). Patients may have very mild skin disease with severe arthritis, or vice versa (6). PsA is challenging to treat and many patients experience progressive joint disease, impaired physical function, and decreased quality of life (3,7,8).

Primary outcomes in trials for immune-mediated diseases often combine multiple relevant measurements into a single composite outcome (9). Because composite measures provide a summary outcome across multiple disease signs and symptoms at a specific point in time, they may be particularly useful for evaluating disease activity and advising treatment decisions in conditions like PsA that have multiple manifestations (2).

Composite measures used to assess PsA include the American College of Rheumatology (ACR) response criteria and Disease Activity Score (28 joints) (DAS28); the Disease Activity [Index] in PsA (DAPSA), which evaluates 66 joints for swelling and 68 joints for tenderness; the PsA Disease Activity Score (PASDAS), which includes the 66/68-joint count as well as the domains of enthesitis and dactylitis; and the Composite Psoriatic Disease Activity Index (CPDAI), which includes enthesitis, dactylitis and skin disease (10). These are continuous measures in which remission is defined as a level below a set cutoff value. Minimal disease activity (MDA) accounts for more than joint involvement, but it is a dichotomous measure representing a state of disease activity (2).

Though several interventional trials have reported PASDAS to be the most sensitive to change among the composite measures, it does not include a formal skin assessment, is complex and difficult to calculate, and is time consuming for both patient and physician (2,11,12). DAPSA is relatively easy to use but also does not assess skin disease (11). Researchers are actively exploring which composite measures have the greatest sensitivity for measuring change in PsA disease activity and the strongest relationship to patient outcomes.

A new composite endpoint in PsA?

Research comparing composite measures has suggested that those measures which include multiple domains are better at quantifying the PsA disease burden (2). They appear to show the greatest sensitivity to change and better represent the breadth of disease manifestations (2). At EULAR 2022, Janssen presented research on a new composite endpoint using pooled data from two Phase 3 clinical trials in active PsA (11). The study suggests that this novel composite endpoint – combining DAPSA low disease activity (LDA; score ≤14, including remission) and Investigator Global Assessment of psoriasis score ≤1 (range=0 [clear] to 4 [severe]) – may be a predictor of long-term PsA skin and joint response, and more practical to implement than PASDAS LDA (11).

Our research into novel composite endpoints is just one of the many avenues we are pursuing to enable more reliable and practical assessments of the complexities of PsA. Janssen is applying machine learning models to help predict treatment response, better understand the molecular causes of disease, and facilitate more accurate and earlier diagnosis. Each of these factors are critically important to making precision medicine a reality for highly variable disease like PsA.

Machine learning unlocks new insights into a key composite endpoint

With PsA holding such diverse clinical phenotypes, machine learning may be useful in identifying potential predictors of response to treatment. Machine learning, a type of artificial intelligence, allows us to detect patterns from vast amounts of clinical data and is already playing an important role toward the development of precision medicine in immunology (13-15). Precision medicine is increasingly significant for heterogeneous diseases like PsA in which a one-size-fits-all approach to treatment may not work for patients with the same diagnosis (14).

In another Janssen study presented at EULAR, researchers described a machine learning analysis of MDA responses using pooled data from biologic-naïve patients with active PsA across two Phase 3 trials (16). MDA is a multi-domain composite and a clinically relevant measure of therapeutic response in PsA, though response patterns in the individual domains have not been well understood (16). MDA domains include tender joint count, swollen joint count, Psoriasis Area and Severity Index and Leeds Enthesitis Index (each ≤1); patient global assessment visual analogue scale (VAS) ≤20; patient pain VAS ≤15; and Health Assessment Questionnaire-Disability Index ≤0.5.16.

The research team sought to characterize treatment response patterns of these domains over time and to identify potential baseline response predictors to therapy. Machine learning identified clusters of PsA patients based on differing response patterns in individual MDA domains (16). These treatment response patterns can further guide researchers and clinicians as we seek to identify the most helpful therapies for individual patient types (16).

Applying novel approaches to the clinical study of chronic inflammatory diseases like PsA shows promising results and great potential for precision medicine applications in clinical research and practice (14).

Since we’re on LinkedIn, I invite you to explore our career opportunities in immunology at Janssen: https://bit.ly/2WKDJLd  

References:

1.       Fransen, J., van Riel, P.L. Outcome measures in inflammatory rheumatic diseases. Arthritis Res Ther. 2009;11(5):244. https://doi.org/10.1186/ar2745  

2.       Coates, L.C. et al. Performance of composite measures used in a trial of etanercept and methotrexate as monotherapy or in combination in psoriatic arthritis. Rheumatology (Oxford). 2021 Mar 2;60(3):1137-1147. https://doi.org/10.1093/rheumatology/keaa271

3.       Belasco, J. & Wei, N. Psoriatic Arthritis: What is Happening at the Joint? Rheumatology and Therapy, 2019: 6(3), 305–315. https://doi.org/10.1007/s40744-019-0159-1

4.       Donvito, T. CreakyJoints: What Is Enthesitis? The Painful Arthritis Symptom You Should Know About. Available at: https://creakyjoints.org/symptoms/what-is-enthesitis/. Accessed May 2022.

5.       Donvito, T. CreakyJoints: What Is Dactylitis? The ‘Sausage Finger’ Swelling You Should Know About. Available at: https://creakyjoints.org/symptoms/what-is-dactylitis/. Accessed May 2022.

6.       Menter, A. Psoriasis and psoriatic arthritis overview. Am J Manag Care. 2016 Jun;22(8 Suppl):s216-24. https://www.ajmc.com/view/psoriasis-and-psoriatic-arthritis-overview

7.       Ogdie, A., Weiss, P. The Epidemiology of Psoriatic Arthritis. Rheum Dis Clin North Am. 2015;41(4):545-568. https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4610151/

8.       Tintle, S.J., Gottlieb, A.B. Psoriatic arthritis for the dermatologist. Dermatol Clin. 2015 Jan;33(1):127-48. https://doi.org/10.1016/j.det.2014.09.010

9.       Grayling, M.J. et al. Innovative trial approaches in immune-mediated inflammatory diseases: current use and future potential. BMC Rheumatol. 2021 Jul 2;5(1):21. https://doi.org/10.1186/s41927-021-00192-5

10.    Duarte-García, A. et al. Endorsement of the 66/68 Joint Count for the Measurement of Musculoskeletal Disease Activity: OMERACT 2018 Psoriatic Arthritis Workshop Report. J Rheumatol. 2019;46(8):996-1005. https://doi.org/10.3899/jrheum.181089

11.    Boehncke, W-H. et al. A Novel Psoriatic Arthritis Composite Endpoint Combining Treatment Targets for Skin and Joints. Presented at EULAR 2022, June 1-4. POS0082.

12.    Mulder, M.L.M. et al. Implementing Psoriatic Arthritis Disease Activity Score-guided treat-to-target in psoriatic arthritis routine clinical practice: (im)possible? Rheumatology (Oxford). 2019 Dec 1;58(12):2330-2331. https://doi.org/10.1093/rheumatology/kez254

13.    Watson, D.S. et al. Clinical applications of machine learning algorithms: beyond the black box. BMJ 2019;364:l886. https://doi.org/10.1136/bmj.l886

14.    Peng, J. et al. Machine Learning Techniques for Personalised Medicine Approaches in Immune-Mediated Chronic Inflammatory Diseases: Applications and Challenges. Front Pharmacol. 2021 Sep 30;12:720694. https://doi.org/10.3389/fphar.2021.720694

15.    Bragazzi N.L. et al. Harnessing Big Data, Smart and Digital Technologies and Artificial Intelligence for Preventing, Early Intercepting, Managing, and Treating Psoriatic Arthritis: Insights From a Systematic Review of the Literature. Front Immunol. 2022 Mar 10;13:847312. https://doi.org/10.3389/fimmu.2022.847312

16.    Zabotti, A. et al. Minimal disease activity response patterns in bio-naïve patients: A machine learning analysis. Presented at EULAR 2022, June 1-4. OP0259.