BACKGROUND: Lupus erythematosus (LE) is a spectrum of autoimmune diseases. Due to the complexity of cutaneous LE (CLE), clinical skin image-based artificial intelligence is still experiencing difficulties in distinguishing subtypes of LE. OBJECTIVES: We aim to develop a multimodal deep learning system (MMDLS) for human-AI collaboration in diagnosis of LE subtypes. METHODS: This is a multi-centre study based on 25 institutions across China to assist in diagnosis of LE subtypes, other eight similar skin diseases and healthy subjects. In total, 446 cases with 800 clinical skin images, 3786 multicolor-immunohistochemistry (multi-IHC) images and clinical data were collected, and EfficientNet-B3 and ResNet-18 were utilized in this study. RESULTS: In the multi-classification task, the overall performance of MMDLS on 13 skin conditions is much higher than single or dual modals (Sen = 0.8288, Spe = 0.9852, Pre = 0.8518, AUC = 0.9844). Further, the MMDLS-based diagnostic-support help improves the accuracy of dermatologists from 66.88% ± 6.94% to 81.25% ± 4.23% (p = 0.0004). CONCLUSIONS: These results highlight the benefit of human-MMDLS collaborated framework in telemedicine by assisting dermatologists and rheumatologists in the differential diagnosis of LE subtypes and similar skin diseases.
Lupus erythematosus (LE) is a heterogeneous, antibody-mediated autoimmune disease. Isolate discoid LE (IDLE) and systematic LE (SLE) are traditionally regarded as the two ends of the spectrum, ranging from skin-limited damage to life-threatening multi-organ involvement. Both belong to LE, but IDLE and SLE differ in appearance of skin lesions, autoantibody panels, pathological changes, treatments, and immunopathogenesis. Is discoid lupus truly a form of LE or is it a completely separate entity? This question has not been fully elucidated. We compared the clinical data of IDLE and SLE from our center, applied multi-omics technology, such as immune repertoire sequencing, high-resolution HLA alleles sequencing and multi-spectrum pathological system to explore cellular and molecular phenotypes in skin and peripheral blood from LE patients. Based on the data from 136 LE patients from 8 hospitals in China, we observed higher damage scores and fewer LE specific autoantibodies in IDLE than SLE patients, more uCDR3 sharing between PBMCs and skin lesion from SLE than IDLE patients, elevated diversity of V-J recombination in IDLE skin lesion and SLE PBMCs, increased SHM frequency and class switch ratio in IDLE skin lesion, decreased SHM frequency but increased class switch ratio in SLE PBMCs, HLA-DRB1*03:01:01:01, HLA-B*58:01:01:01, HLA-C*03:02:02:01, and HLA-DQB1*02:01:01:01 positively associated with SLE patients, and expanded Tfh-like cells with ectopic germinal center structures in IDLE skin lesions. These findings suggest a significant difference in the immunopathogenesis of skin lesions between SLE and IDLE patients. SLE is a B cell-predominate systemic immune disorder, while IDLE appears limited to the skin. Our findings provide novel insights into the pathogenesis of IDLE and other types of LE, which may direct more accurate diagnosis and novel therapeutic strategies.
