Study on Clinical Prediction Models for the Risk of Malignant Tumors in Patients with Systemic Lupus Erythematosus (Preprint)

BACKGROUND Systemic lupus erythematosus (SLE) is a multi-organ autoimmune disease. Long-term use of immunosuppressive drugs in patients can lead to cellular DNA damage, increasing the risk of developing cancer. 1-3 Studies have shown that SLE patients will develop solid tumors such as lung, breast, bladder, uterus, and hematologic malignant tumor complications such as non-Hodgkin lymphoma within 8~15 years after the onset of the disease. The coexistence of malignancies is one of the leading causes of mortality in SLE patients. 4-7 Therefore, reducing the risk of cancer in patients is crucial for prolonging their lifespan. Currently, clinical prevention of malignant tumor complications in SLE patients involves regular tissue examinations, which do not effectively achieve early prevention. While constructing disease prediction models could forecast patient risk at an earlier stage for early prevention, there is currently a lack of risk prediction models for malignant tumors in SLE patients. 8 This study aims to construct a risk prediction model for malignant tumors in SLE patients by collecting relevant clinical indicators, to enable early prediction and prevention of malignant tumors in SLE patients, thereby improving their survival rate. OBJECTIVE Collect and record the signs and relevant laboratory data of systemic lupus erythematosus (SLE) patients from Affiliated Hospital of North Sichuan Medical University and Bazhong Central Hospital, and establish a risk prediction model for malignancy complicating SLE patients. METHODS We selected 504 SLE patients diagnosed and treated at the Affiliated Hospital of North Sichuan Medical University and Bazhong Central Hospital, from January 2016 to January 2023, including 22 patients with concurrent malignancy. Record whether the patient has the presence of Raynaud's, photosensitivity, arthralgia, fever, mouth ulcer, and alopecia. Laboratory data including urinary protein, globulin, albumin, LDH, creatinine, lymphocyte, platelets, neutrophils, monocytes, ESR, CRP, IgA, IgG, IgM, IgE, C3, C4, CEA, CA199, CA125, CA153, AFP, SSA, SSB, dsDNA, Sm antibodies, anti-nucleolar antibodies, anti-histone antibodies, anti-ribosomal P protein antibodies, and anti-phospholipid antibodies were collected. Analyze the risk factors affecting malignancy in SLE patients using Logistic regression, and based on the results, construct a predictive model for malignancy in SLE patients using ROC curves. RESULTS Compared with SLE patients without malignancy, SLE patients with malignancy showed significant differences in age, Raynaud, ESR, CRP, IgG, C4, CEA, CA199, CA125, CA153, AFP, and Sm antibodies (p < 0.05). The following factors were identified as significant risk factors for malignancy in SLE patients: Age (OR=1.04, 95% CI: 1.02-1.07, p=0.002), Fever (OR=4.29, 95% CI: 1.76-10.42, p=0.001), mouth ulcer (OR=2.43, 95% CI: 1.01-5.83, p=0.048), Lymphocyte count (OR=0.16, 95% CI: 0.05-0.57, p=0.005), ESR (OR=1.02, 95% CI: 1.01-1.04, p < 0.001), CRP (OR=1.02, 95% CI: 1.00-1.03, p=0.006), C4 (OR=1.01, 95% CI: 1.01-1.02, p < 0.001), CEA (OR=74.51, 95% CI: 17.53-316.65, p < 0.001)； CA199 (OR=5.73, 95% CI: 1.76-18.67, p=0.004), CA125 (OR=4.65, 95% CI: 1.80-12.00, p=0.002), Sm antibody (OR=5.73, 95% CI: 2.20-14.94, p < 0.001). The individual AUC values for the above factors were as follows: Age: 0.700, Fever: 0.560, mouth ulcer: 0.608, Lymphocyte count: 0.713, ESR: 0.724, CRP: 0.737, C4: 0.740, CEA: 0.671, CA199: 0.569, CA125: 0.608, Sm antibody: 0.689. When these factors were combined to construct a prediction model, the AUC value increased to 0.917, indicating a high predictive capacity. CONCLUSIONS The factors of age, fever, mouth ulcer, lymphocyte count, ESR, CRP, C4, CEA, CA199, CA125, and Sm antibodies can be used to construct a risk prediction model for malignancy in SLE patients.