Human-multimodal deep learning collaboration in 'precise' diagnosis of lupus erythematosus subtypes and similar skin diseases.

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.

Clinical features and prognostic factors of severe cutaneous adverse drug reactions: A single-center retrospective study of 209 cases in China.

BACKGROUND: Severe cutaneous adverse drug reactions (SCAR) are life-threatening and contain drug reactions with eosinophilia and systemic symptoms (DRESS), Stevens-Johnson syndrome (SJS), toxic epidermal necrolysis (TEN), and acute generalized exanthematous pustulosis (AGEP). METHODS: We aimed to evaluate clinical features and prognostic factors for SCAR patients. From January 2010 to April 2022, 209 patients with SCAR (DRESS, n = 46, SJS/TEN, n = 128, AGEP, n = 35) were included in this study. Clinical symptoms, laboratory tests, causative drugs, disease courses, treatments, and outcomes were investigated. RESULTS: Antibiotics ranked first (35.9 %) followed by traditional Chinese medicine (15.8 %) and antiepileptic drugs (14.8 %) among causative drugs of SCAR. One patient (2.2 %) with DRESS and seven patients (5.5 %) with SJS/TEN died in the hospital, while there was no AGEP-related mortality. The multivariate logistic regression analysis showed that high Registry of Severe Cutaneous Adverse Reactions score (OR = 2.340, 95 % CI = 1.192-4.591) and hemoglobin < 100 g/L (OR = 0.126, 95 % CI = 0.016-0.983) were independent risk factors of DRESS. Anemia (OR = 0.191, 95 % CI = 0.037-0.984) and body surface area detached involved at day 1 (OR = 2.749, 95 % CI = 1.115-6.778) were independent risk factors of SJS/TEN for severe acute complications and hospital death (P < 0.05). Lymphocytopenia (OR = 0.004, 95 % CI = 0.000-0.553) was a risk factor of AGEP for acute complications (P = 0.028). CONCLUSION: This study reveals the clinical features and independent prognostic factors for SCAR, which may be helpful in the clinical management for SCAR patients.

DNA methylation markers in peripheral blood for psoriatic arthritis.

BACKGROUND: The clinical manifestations of psoriatic arthritis (PsA) are highly heterogeneous and no reliable diagnostic biomarkers exist. OBJECTIVE: We explored the role of DNA methylation CpG markers in the diagnosis of PsA. METHODS: DNA methylation array was used to screen for differentially methylated sites (DMSs) in the discovery phase (PsA, n = 25; healthy controls [HCs], n = 19; psoriasis vulgaris [PsV], n = 20). In the validation phase, pyrosequencing was used to identify the DMSs in an expanded cohort (PsA, n = 60; HCs, n = 91; PsV, n = 48; rheumatoid arthritis [RA], n = 60). Logistic regression prediction models were established based on the identified DMSs for the diagnosis of PsA. RESULTS: A total of 17 DMSs differentiating PsA and HCs as well as 11 DMSs differentiating PsA and PsV were screened in the discovery phase. A total of six DMSs (chr14: cg07940072, chr14: 38061320, chr9: cg15734589, chr6: cg12800266, chr3: cg12992827, chr6: cg24500972) differentiating PsA and HCs and two DMSs (chr12: cg16459382, chr2: cg16348668) differentiating PsA and PsV were identified using pyrosequencing. Three logistic regression prediction models were established based on the identified DMSs, which distinguished PsA, RA, PsV, and HCs (P < 0.001). The models performed well in differentiating PsA from HCs, RA, and PsV (AUC: 0.858, 0.851, and 0.976, respectively). CONCLUSIONS: The models based on methylated CpG sites are useful for distinguishing patients with PsA from HCs and those with RA or PsV and are a highly sensitive and specific diagnostic biomarker for PsA.

Decreased microRNA-126 expression in psoriatic CD4+ T cells promotes T-helper 17 cell differentiation and the formation of dermatitis in imiquimod-induced psoriasis-like mice.

Psoriasis is a chronic inflammatory skin disease with multiple genetic backgrounds, whose etiology and pathogenesis are still unclear. Complex T-cell immune imbalance has been demonstrated to play an important role in pathogenesis of psoriasis. This study reported that microRNA-126 (miR-126) expression was decreased in CD4+ T cells of both psoriasis patients and psoriasis-like mouse models and its expression was negatively correlated with the Psoriasis Area and Severity Index (PASI) score of psoriasis patients. Conditional Mir126 knockout in mouse CD4+ T cells can obviously aggravate the psoriasis-like dermatitis and promote T-helper (Th)1 and Th17 cells' infiltration in spleen of imiquimod (IMQ)-induced psoriasis-like mouse model. In addition, the mRNA expression of Il17a and Il17f were significantly increased in mouse naïve CD4+ T cells with Mir126 knockout after stimulating with CD3 and CD28. Compared with naïve CD4+ T cells, the expression of Mir126 was decreased in Th17 cells, and Mir126 knockout notably promoted the differentiation of naïve CD4+ T cells to Th17 cells as well as the mRNA expression of Il17a, Il17f, Rorc, and Il23R. Our results revealed that decreased miR-126 in psoriatic CD4+ T cells might accelerate the formation of skin lesions through promoting the differentiation of Th17 cells, thus suggesting a potential intervention target for psoriasis.