Idiopathic inflammatory myopathies often affect extramuscular organs such as the skin, joints, lungs, heart and gastrointestinal tract (1). In particular, interstitial lung disease (ILD) is one of the major manifestations associated with poor mortality in patients with polymyositis (PM)/dermatomyositis (DM) (2). The management of ILD in patients with PM/DM aims to ameliorate, stabilize, or slow its progression based on the disease behavior of ILD (3). In terms of treatment for myositis-associated ILD, systemic corticosteroid therapy is usually combined with immunosuppressive agents, such as azathioprine, cyclophosphamide, mycophenolate mofetil, methotrexate, cyclosporine, tacrolimus, and/or rituximab, although there is little evidence to support the efficacy of these individual agents (4). Since the clinical course, response to treatment, and prognosis are highly variable among patients with myositis-associated ILD (3, 5), the treatment regimen is decided based mainly on the progression speed and severity of ILD. On the other hand, many studies have reported clinical, laboratory, and imaging features predicting subsequent treatment response and prognosis (6–11).

Among myositis-specific autoantibodies, anti-melanoma differentiation-associated gene 5 (MDA5) and anti-aminoacyl tRNA synthetase (ARS) antibodies are associated with ILD (2). Anti-MDA5 antibody is useful for predicting rapidly progressive ILD and poor survival in patients with myositis-associated ILD (9, 12). Approximately 30 to 50% of patients with anti-MDA5-associated ILD die of respiratory failure within 6 months after diagnosis (9, 13, 14). In a large-scale multicenter retrospective cohort of Japanese patients with myositis-associated ILD (JAMI), the major cause of death was respiratory insufficiency directly related to ILD and ant-MDA5 antibody was the strongest predictor of mortality regardless of the initial treatment regimen (9, 11). To overcome this devastating condition, an intensive immunosuppressive regimen consisting of high-dose corticosteroids, a calcineurin inhibitor maintained at a high serum trough level, and intermittent intravenous cyclophosphamide (“triple-combo” therapy) is empirically used mainly in Japan without firm evidence for efficacy (15–17). A prospective, multicenter, single-arm study in patients with anti-MDA5-associated ILD suggested superiority of initial triple-combo therapy over historical controls who received sequential combination therapy (18). On the other hand, a retrospective cohort reported that more than half of patients with myositis-associated ILD who received triple-combo therapy developed serious infection events, including bacterial, fungal and cytomegalovirus infection (19). In addition, it has been reported that some patients with anti-MDA5-associated ILD favorably respond to corticosteroids combined with a single immunosuppressive agent (“dual-combo” therapy) or corticosteroids alone (monotherapy) (20). In fact, over 80% of patients with anti-MDA5 antibody survived without receiving initial triple-combo therapy in the JAMI cohort (our unpublished data). On the other hand, patients with myositis-associated ILD who were negative for anti-MDA5 antibody are sometimes resistant to initial dual-combo therapy (11); initial triple-combo therapy might have a role in this patient population. Therefore, a personalized approach is necessary to optimize the initial immunosuppressive regimen in patients with myositis-associated ILD, especially those with anti-MDA5-associated ILD. In this study, the JAMI cohort was used to stratify patients with myositis-associated ILD who were initially treated with triple-combo therapy into subgroups based on mortality outcomes by a cluster analysis.

We successfully developed a clustering model that predicts survival in patients with myositis-associated ILD who received initial triple-combo therapy. The patient subsets categorized by optimal clustering differed with regard to age at disease onset, disease classification, autoantibodies, serum biomarkers, and severity of ILD, which are known as predicting factors for mortality in patients with myositis-associated ILD (6–11). Our findings suggest that subgrouping of patients with myositis-associated ILD based solely on the presence or absence of anti-MDA5 antibody may not capture the heterogeneous treatment responses to the intensive immunosuppressive regimen. This model enables us to predict the survival of patients with myositis-associated ILD undergoing triple-combo therapy but fails to identify patients who benefit from initial triple-combo therapy over dual-combo therapy or monotherapy. It was of note that prognosis in patients with myositis-associated ILD correlated with baseline characteristics, such as autoantibodies profiles and ILD severity, rather than the intensity of initial immunosuppressive treatment.

Triple-combo therapy has often been used for myositis patients who develop acute/subacute ILD, particularly those with anti-MDA5 antibody (15–17). Survival rates in patients with anti-MDA5-associated ILD who received triple-combo therapy ranged from 42 to 85% according to previous reports (16, 18, 27, 28). Our cluster model was able to classify anti-MDA5-positive patients who received triple-combo therapy primarily into two subsets based on prognosis: a subset with moderate mortality rates (clusters #3 and #4) and a subset with high mortality rates (cluster #6). In addition, a minor population of anti-MDA5-positive patients was included in the low mortality group (cluster #1). Cluster #6, with poor treatment responses to triple-combo therapy, was characterized by older age at disease onset, increased levels of CRP and KL-6, and hypoxia at diagnosis, which were reported as poor prognostic factors in patients with myositis-associated ILD (9, 11). These patients probably require novel treatment approaches, such as the use of a Janus kinase inhibitor and plasma exchange (29–36). However, no survival benefit of triple-combo therapy over dual-combo therapy or monotherapy was shown in clusters #3, #4, or #6. In addition, our cluster model failed to identify patients with anti-MDA5-associated ILD who did not require triple-combo therapy. Finally, our findings did not support a better prognosis in patients with anti-MDA5-associated ILD treated with initial triple-combo therapy over those treated initially with dual-combo therapy or monotherapy (18). This finding was a rather unexpected. The precise reason for considerable difference in the mortality between anti-MDA5 antibody-positive patients undergoing triple-combo therapy in the study by Tsuji et al. (18) and those in the JAMI cohort was unclear, but this might be attributable to the different criteria for inclusion. In the study by Tsuji et al. (18), all consecutive patients identified to be positive for anti-MDA5 antibody received initial triple-combo therapy irrespective of the prognostic factors. On the other hand, in the JAMI cohort, anti-MDA5 antibody was measured using the stored serum samples later in the central laboratories and attending physicians had to decide initial treatment regimen without knowing the presence or absence of anti-MDA5 antibody. Therefore, in the JAMI cohort, triple combo therapy was chosen based on physician's expertise, not on information of the presence of anti-MDA5 antibody. Actually, 28% of the patients in the triple-combo therapy group were anti-MDA5 antibody-negative. The patients with more severe ILD or those with multiple prognostic factors for poor outcomes were likely to be included in the triple-combo therapy group in the JAMI cohort.

On the other hand, the efficacy of triple-combo therapy in patients negative for anti-MDA5 antibody remains uncertain. In our clustering model, clusters #1, #2, and #5 mainly consisted of anti-MDA5 antibody-negative patients. According to previous reports on anti-MDA5 antibody-negative patients with myositis-associated ILD, the cumulative survival rate over 5 years was >80%, regardless of the initial immunosuppressive regimen (12, 37–39). Our cluster model identified patients who had worse survival despite treatment with triple-combo therapy among anti-MDA5 antibody-negative patients with myositis-associated ILD (cluster #5). The cluster #5 consisted mainly of elderly patients with high levels of CRP and KL-6 who required supplemental oxygen at diagnosis. Patients with such features at presentation need to be managed with caution even in the absence of anti-MDA5 antibody. On the other hand, in anti-MDA5 antibody-negative patients categorized into clusters #1 and #2, there was no survival benefit of triple-combo therapy over dual-combo therapy or monotherapy.

Unexpectedly, in clusters #3 and #4, survival rates were better in the dual-combo therapy or monotherapy group than in the triple-combo therapy group. This finding needs to be carefully interpreted, and could be potentially due to inadequate adaptation of the clustering model developed for patients treated with triple-combo therapy to those treated with dual-combo therapy or monotherapy. In fact, clinical characteristics were somewhat different between the triple-combo therapy group and the dual-combo therapy or monotherapy group in individual clusters. Since the treatment regimen was decided by attending physicians based on their expertise without information of autoantibody profiles in the JAMI cohort, the JAMI dataset might not be able to capture all relevant factors related to the physician's expertise for selection of the initial treatment regimen, such as patient report outcomes and physician's global estimate of patient status.

There are several limitations in this study. First, the participating centers of the JAMI cohort consist mainly of tertiary referral hospitals, which are likely to enroll patients with more severe disease, such as those with anti-MDA5 antibody. Second, the JAMI cohort did not enroll patients with anti-ARS antibody without any muscle or skin symptoms, which are often encountered in clinical practice. Third, pulmonary function test and arterial blood gas data at diagnosis were missing in considerable proportions of patients enrolled in the JAMI cohort, while baseline forced vital capacity was identified as the prognostic factor for mortality in anti-MDA5 antibody-positive patients with DM-associated ILD (40). Finally, we used initial predictors for all-cause mortality or mortality due to respiratory insufficiency directly related to ILD reported in previous reports using the JAMI cohort (9, 11) as variables in the cluster analysis, although there are many other prognostic factors for mortality in patients with myositis-associated ILD reported previously (6–8, 10, 39, 40).

We successfully developed a cluster model that categorizes patients with myositis-associated ILD who were treated with initial triple-combo therapy into subgroups with different prognoses. However, the patient subgroup that benefitted from triple-combo therapy was not identified in this cluster analysis.

The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

The studies involving human participants were reviewed and approved by Ethics Committee of Nippon Medical School. The patients/participants provided their written informed consent to participate in this study.

Naoshi Nishina and Yuko Kaneko, Keio University School of Medicine, Tokyo, Japan; Yasushi Kawaguchi, Tokyo Women's Medical University, Tokyo, Japan; Atsushi Kawakami, Nagasaki University Graduate School of Biomedical Sciences, Nagasaki, Japan; Kei Ikeda, Chiba University Hospital, Chiba, Japan; Yohei Kirino and Yukie Yamaguchi, Yokohama City University Graduate School of Medicine, Yokohama, Japan; Yumiko Sugiyama, Yokohama City University Medical Center, Yokohama, Japan; Yoshinori Tanino, Fukushima Medical University School of Medicine, Fukushima, Japan; Takahiro Nunokawa, Tokyo Metropolitan Tama Medical Center, Fuchu, Japan; Katsuaki Asakawa, Niigata University Medical and Dental Hospital, Niigata, Japan; Taro Ukichi, The Jikei University School of Medicine, Tokyo, Japan; Shinjiro Kaieda, Kurume University School of Medicine, Kurume, Japan; Taio Naniwa, Nagoya City University School of Medicine, Nagoya, Japan; Yutaka Okano, Kawasaki Municipal Hospital, Kawasaki, Japan; Yoshinori Taniguchi, Kochi Medical School, Kochi University, Kochi, Japan; Jun Kikuchi, Saitama Medical Center, Saitama Medical University, Saitama, Japan; Makoto Kubo, Yamaguchi University Graduate School of Medicine, Yamaguchi, Japan, Masaki Watanabe, Graduate School of Medical and Dental Sciences, Kagoshima University, Kagoshima, Japan; Tatsuhiko Harada, Nagasaki University School of Medicine, Nagasaki, Japan; Taisuke Kazuyori, The Jikei University School of Medicine Katsushika Medical Center, Tokyo, Japan; Hideto Kameda, Toho University Ohashi Medical Center, Tokyo, Japan; Makoto Kaburaki, Toho University School of Medicine, Tokyo, Japan; Yasuo Matsuzawa, Toho University Medical Center, Sakura Hospital, Tokyo, Japan; Shunji Yoshida, Fujita Health University School of Medicine, Toyoake, Japan; Yasuko Yoshioka and Takuya Hirai, Juntendo University Urayasu Hospital, Chiba, Japan; Yoko Wada, Niigata University Graduate School of Medical and Dental Sciences, Niigata, Japan; Koji Ishii, Faculty of Medicine, Oita University, Oita, Japan; Sakuhei Fujiwara, Faculty of Medicine Oita University, Oita, Japan; Takeshi Saraya, Kyorin University, Mitaka, Japan; Kozo Morimoto, Fukujuji Hospital, Japan Anti-Tuberculosis Association, Kiyose, Japan; Tetsu Hara, Hiratsuka Kyosai Hospital, Hiratsuka, Japan; Hiroki Suzuki, Saiseikai Yamagata Saisei Hospital, Yamagata, Japan; Hideki Shibuya, Tokyo Teishin Hospital, Tokyo, Japan; Yoshinao Muro, Nagoya University Graduate School of Medicine, Nagoya, Japan; Ryoichi Aki, Kitasato University School of Medicine, Sagamihara, Japan; Takuo Shibayama, National Hospital Organization Okayama Medical Center, Okayama, Japan; Shiro Ohshima, National Hospital Organization Osaka Minami Medical Center, Kawachinagano, Japan; Yuko Yasuda, Saiseikai Kumamoto Hospital, Kumamoto, Japan; Masaki Terada, Saiseikai Niigata Daini Hospital, Niigata, Japan; Yoshie Kawahara, Keiyu Hospital, Yokohama, Japan.

TG, KM, SS, and MK: study conception and design. TG, KM, SS, MK, and JAMI Investigators: acquisition of data. TG and KM: analysis and interpretation of data. All authors draft the article or revise it critically for important intellectual content. All authors contributed to the article and approved the submitted version.

This work was supported in part by a research grant for intractable diseases from the Japanese Ministry of Health, Labour and Welfare (20FC1050), Japan Agency for Medical Research and Development (21ek0109531h0001), and a research grant from Astellas. The funders had no role in the study design, data collection and analysis, decision to publish, or preparation of the manuscript.

TG received speaking fees from Astellas, Boehringer Ingelheim, Bristol-Myers Squibb, Janssen, MBL, Nippon Shinyaku, and Ono Pharmaceuticals. SS holds the patent for the anti-MDA5 antibody measurement kit. MK holds the patent for the anti-MDA5 antibody measurement kit and received consulting fees, speaking fees, and research grants from AbbVie, Actelion, Asahi Kasei, Astellas, Boehringer Ingelheim, Bayer, Chugai, Eisai, Corbus, Janssen, Kissei, MBL, Mitsubishi Tanabe, Mochida, Nippon Shinyaku, Pfizer, Ono Pharmaceuticals, Reata, and Teijin. The remaining author declares that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.