Lupus nephritis (LN) is a main manifestation of systemic lupus erythematosus (SLE), which will largely effect the prognosis of SLE patients. Our previous 10-year data showed that the development of LN is most common in the first year of SLE, occupying about 17%. And our group has established a prediction model to predict the 1-year probability of LN for SLE patients without renal involvement. Our previous proof-of-concept trial and multicenter, randomized, double-blind, placebo-controlled trial indicated that metformin seemed to have potential to reduce the new-onset of LN in SLE patients (Unpublished data, in review). So the investigators tried to illustrate whether metformin has effect to prevent the development of lupus nephritis in high risk SLE patients based on LN prediction model.

Inclusion Criteria:

• New-onset of SLE (within 1 year)
• Meet the 1997 ACR criteria of SLE
• 1-year risk of developing LN higher than 50% based on LN prediction model.
• All the patients received standard care of therapy. (mycophenolate/azathioprine/cyclosporin/tacrolimus/methotrexate/leflunomide were allowed).
• Sign the informed consent.

Exclusion Criteria:

• Patients with kidney involvement at screening were excluded
• Patients with severe SLE were excluded, such as neuropsychiatric and cardiac involvement
• Cyclophosphamide or biologics were not allowed.
• Alanine aminotransferase (ALT)/ aspartate aminotransferase (AST) > 2 times upper normal limits; creatinine clearance rate < 60ml/min;
• Previous exposure of metformin within 30 days before screening; or previous history of intolerant to metformin;
• Patients who diagnosed of diabetes mellitus;
• Pregnancy or lactation.