1047: A Novel Prediction Tool for Progression of Systemic Sclerosis-Interstitial Lung Disease Despite Treatment with Immunosuppression

Elizabeth Volkmann¹, Holly Wilhalme¹, Grace Kim¹, Jonathan Goldin¹, Shervin Assassi², Masataka Kuwana³, Donald Tashkin¹ and Michael roth¹, ¹University of California Los Angeles, Los Angeles, CA, ²McGovern Medical School, University of Texas, Houston, TX, ³Nippon Medical School Graduate School of Medicine, Tokyo, Japan

Background/Purpose: Systemic sclerosis-interstitial lung disease (SSc-ILD) is the leading cause of SSc-related death. While some patients respond favorably to treatment with immunosuppression, a subset of patients will experience ILD progression despite treatment. Early loss of lung function, within the first 2 years of treatment, is an independent predictor of mortality. Currently, no evidence-based tools exist to identify patients with a progressive SSc-ILD phenotype. The purpose of this study was to create a prediction tool to identify those patients who are less likely to experience an improvement in ILD in response to treatment with mycophenolate (MMF) or cyclophosphamide (CYC).

Methods: Participants of Scleroderma Lung Study (SLS) II¹ were included in this analysis. SLS II randomized patients with SSc-ILD to 2 years of MMF or 1 year of CYC, followed by 1 year of placebo. ILD progression was defined as a >=2% increase in the quantitative extent of ILD in the whole lung (QILD-WL) on HRCT of the chest at 2 years. This radiographic threshold of ILD worsening predicted long-term mortality in both SLS II and SLS I.² Univariate logistic regression analysis was used to identify baseline variables associated with ILD progression. Baseline variables included demographic characteristics, treatment arm, SSc disease features, ILD severity, reflux severity, and a focused group of serum proteins and auto-antibodies previously found to correlate with extent/progression of SSc-ILD. Those variables associated with outcome (P < 0.1) were then combined into a multivariable model. Receiver-operating characteristic (ROC) analysis was performed. The Bootstrap method was applied to internally validate the model.

Results: Among SLS II participants who had a follow-up HRCT at 2 years (N=97), 29% experienced radiographic progression of ILD (change in QILD-WL >=2%). There were no significant differences between baseline characteristics of participants with and without the follow up HRCT in SLS II. In the univariate analysis, increased baseline reflux score (P=0.02), increased baseline KL-6 level (P=0.02) and male sex (P=0.05) were associated with worsening of QILD-WL >=2%. The area under the ROC curve for the multivariable model was 0.71 (Table 1; Figure 1). A Forrest plot demonstrates the odds ratios for the individual variables (Figure 2).

Conclusion: A clinical prediction tool defined by male sex, increased KL-6, increased reflux severity had good discrimination for predicting progression of SSc-ILD despite treatment with mycophenolate or cyclophosphamide. Patients with this SSc-ILD phenotype may benefit from a more aggressive initial treatment approach and closer monitoring. This prediction tool requires validation in other cohorts. 

References:
1. Tashkin DP, et al. Lancet Resp Med 2016;4:708.
2. Volkmann ER, et al. Chest 2022;161:P1310.


Figure 1. ROC curve for predicting radiographic progression of SSc-ILD

Figure 1. Forest plot of odds ratios and 95% confidence intervals for multivariable model predicting SSc-ILD progression. Odds ratios greater than 1 indicate increased risk of progression (higher KL-6 and higher reflux scores); odds ratios less than 1 indicate decreased risk of progression (female sex).
Disclosures: E. Volkmann, Boehringer-Ingelheim, Kadmon, Horizon; H. Wilhalme, None; G. Kim, University of California Los Angeles, MedQIA, Boehringer-Ingelheim; J. Goldin, MedQIA, LLC; S. Assassi, Boehringer-Ingelheim, Janssen, Novartis, AstraZeneca, CSL Behring, AbbVie/Abbott; M. Kuwana, Boehringer-Ingelheim, Ono pharmaceuticals, Mochida, AbbVie/Abbott, Astellas, Janssen, Bayer, Corbus, Horizon; D. Tashkin, Genentech; M. roth, Genentech.