Background/Purpose: Develop predictive models of early (8 week) and late (24 week) disease outcomes using clinical features, autoantibodies, and serum biomarkers in patients with refractory myositis treated with rituximab.

Methods: In the Rituximab in Myositis (RIM) trial, all subjects (76 with adult dermatomyositis, 76 with adult polymyositis and 48 with juvenile dermatomyositis) received rituximab (2 doses on consecutive weeks) with half the patients receiving drug at baseline and half receiving drug 8 weeks later. Using start of treatment as baseline, serum samples (n=177) were analyzed at baseline and after 8 and 24 weeks after rituximab. Potential predictors included the following baseline factors: clinical features, serum muscle enzymes, interferon gene score, autoantibodies (anti-synthetase n=28, TIF1-g n=19, Mi-2 n=25, SRP n=21, NXP2 n=18, non-myositis associated n=24, undefined autoantibody n=9), and cytokines/chemokines measured by multiplexed sandwich immunoassays (Meso Scale Discovery) (type-1 IFN-inducible [IP-10, I-TAC, MCP1], Th1 [IFNγ, TNFα, IL2], Th2 [IL4, IL5, IL10, IL12, IL13], Th17 [IL6, IL17, IL1β] and regulatory cytokines [IL10, TNFa, MIP-1α, MIP-1β]). Our primary definition of response to treatment was based on absolute change from baseline to 8 weeks and 24 weeks in physician global visual analog scale (VAS), muscle VAS, and extramuscular VAS. Multivariable linear regression models were developed using stepwise variable selection methods.

Results: Preliminary models were built with good predictive ability both for change in physician global assessment and muscle disease activity at 24 weeks (R-square=0.41 and 0.40, respectively). The model for change in physician global assessment included the following baseline clinical and lab features: muscle disease activity, physician global assessment, and I-TAC (Table). The model for change in muscle disease activity included baseline physician global assessment, skeletal disease activity, I-TAC and IFNγ. Similarly, a predictive model was built with excellent predictive ability (R-square=0.67) for change in extramuscular disease activity at 24 weeks. This model included the following baseline clinical and lab features: constitutional, skeletal and extramuscular disease activity by VAS, and MIP-1β and Mi-2. We also built models from baseline to 8 weeks but their predictive ability was inferior compared to those for 24 weeks (R-square<0.3).

Conclusion: Changes in disease activity over time following treatment with rituximab in patients with refractory myositis can be predicted.  These models could be clinically useful to optimize treatment selection in these patients.

Outcomes ->	Change in Physician Global VAS		Change in Muscle VAS		Change in Extramuscular VAS
Predictors (baseline)	Coefficient	P-value	Coefficient	P-value	Coefficient	P-value
Physician Global Assessment	-0.80	0.0007	-0.33	0.002	—	—
Muscle Disease Activity	0.51	0.0008	—	—	—	—
Extramuscular Global Assessment	—	—	—	—	0.60	0.0002
Skeletal Disease Activity	—	—	0.38	0.02	0.26	0.04
Constitutional Disease Activity	—	—	—	—	0.23	0.02
Mi-2	—	—	—	—	-15.47	0.002
I-TAC	-0.01	0.025	-0.02	0.01	—	—
IFNγ	—	—	-0.35	0.01	—	—
MIP-1β	—	—	—	—	0.008	<0.0001

---

« Back to 2014 ACR/ARHP Annual Meeting

ACR Meeting Abstracts - https://acrabstracts.org/abstract/a-predictive-model-of-disease-outcome-in-rituximab-treated-myositis-patients-using-clinical-features-autoantibodies-and-serum-biomarkers/