Background Polymyositis (PM), dermatomyositis (DM) and clinically amyopathic DM (CADM) are autoimmune myositis which can be associated with interstitial lung disease (ILD).^1,2 The relapse rate of ILD is high, reported as approximately 20%–55%.^3,4 Since relapses result in decreased pulmonary function, it is important to identify the predictive factors for the relapse.

Objectives The aim of this study was to elucidate the predictive factors for the relapse of ILD associated with myositis (PM/DM/CADM).

Methods We conducted an observational retrospective study. Patients with myositis-associated ILD who have ever visited our institution between 2002–2017 and achieved remission once were enrolled. Patients who died before remission were excluded. We collected their clinical information from medical records. We compared patient characteristics between relapse group and non-relapse group by Fisher’s exact test or Mann-Whitney U test at first. Relapse was defined as exacerbation of radiological findings of which doctor-in-charge decided to intensify therapy for ILD. We performed Kaplan-Meier analysis to compare the relapse-free survival for the characteristics that had significant differences between two groups. To perform Kaplan-Meier analysis, continuous variables were converted to dichotomous variables for analysis by setting cut-off values determined by receiver-operating characteristics (ROC). Then, using each characteristic which showed significant difference in Kaplan-Meier analysis, we conducted Cox’s proportional hazard analysis for multivariate analysis. For relapse group, we examined the changes of serum KL-6 levels from the initial treatment of myositis-associated ILD to the relapse of ILD. We calculated the average of serum KL-6 levels of 3 months and 6 months before relapse, respectively, then compared them with KL-6 levels at the time of relapse.

Results Seventy-two patients with myositis-associated ILD at our institution were enrolled. Among 72 patients, 24 experienced relapse (relapse group) and 48 did not experience relapse (non-relapse group). Median observational period was 31.5 months and 39.0 months, respectively. Median levels of serum KL-6, the rate of patients who had upper lung field (ULF) lesion by CT, and anti-ARS antibody prevalence were significantly higher in relapse group than in non-relapse group (1870 vs 935 U/mL, p=0.003; 62 vs 27%, p=0.01; 88 vs 60%, p=0.03, respectively). Median levels of%VC was significantly lower in relapse group than in non-relapse group (65.7 vs 81.2%, p=0.02). By ROC analysis, the cut-off levels of serum KL-6 and%VC were determined as 1359 U/mL and 70.5%, respectively. Kaplan-Meier analysis showed serum KL-6 >1359 U/mL(p=0.02), anti-ARS antibody (p<0.05),%VC <70.5 (p=0.004), and ULF lesions (p=0.01) were significantly related to the relapse (figure 1). Multivariate analysis revealed only serum KL-6 >1359 U/mL was an independent risk factor for relapse (hazard ratio: 4.9 (95%CI 1.0–24.0), p<0.05) among the 4 characteristics. At the time of the relapse, serum KL-6 levels were increased 37% from the 3 months average and 51% from the 6 months average.

Conclusions Serum KL-6 was a strong predictor for relapse of myositis-associated ILD.

References [1] N Engl J Med. 1975;292:344.

[2] J Am Acad Dermatol. 2002;46:626.

[3] Autoimmunity2006;39:233.

[4] Rheumatology (Oxford)2015;54:39.

Disclosure of Interest None declared