Clinico-sero-pathological profiles and risk prediction model of idiopathic inflammatory myopathy (IIM) patients with different perifascicular changes

Lining Zhang,Lijun Fu,Guoyong Zhang,Ying Hou,Xiaotian Ma,Dandan Zhao,Wei Li,Tingjun Dai,Qiang Shu,Chuanzhu Yan,Bing Zhao
DOI: https://doi.org/10.1111/cns.14882
Abstract:Aims: To explore the clinico-sero-pathological characteristics and risk prediction model of idiopathic inflammatory myopathy (IIM) patients with different muscular perifascicular (PF) changes. Methods: IIM patients in our center were enrolled and the clinico-sero-pathological data were retrospectively analyzed. A decision tree model was established through machine learning. Results: There were 231 IIM patients enrolled, including 53 with perifascicular atrophy (PFA), 39 with perifascicular necrosis (PFN), and 26 with isolated perifascicular enhancement of MHC-I/MHC-II (PF-MHCn). Clinically, PFA patients exhibited skin rashes and dermatomyositis-specific antibodies (DM-MSAs, 74.5%) except for anti-Mi2. PFN patients showed the most severe muscle weakness, highest creatine kinase (CK), anti-Mi2 (56.8%), and anti-Jo-1 (24.3%) antibodies. PF-MHCn patients demonstrated negative MSAs (48.0%) and elevated CK. Histopathologically, MAC predominantly deposited on PF capillaries in PFA but on non-necrotic myofiber in PFN (43.4% and 36.8%, p < 0.001). MxA expression was least in PF-MHCn (36.0% vs. 83.0% vs. 63.2%, p < 0.001). The decision tree model could effectively predict different subgroups, especially PFA and PFN. Conclusions: Three types of PF change of IIMs representing distinct clinico-serological characteristics and pathomechanism. Undiscovered MSAs should be explored especially in PF-MHCn patients. The three pathological features could be accurately predicted through the decision tree model.
What problem does this paper attempt to address?