Systemic lupus erythematosus is a chronic autoimmune disease, which can involve multiple systems and largely impair patients' health. Kidney is the one of the most commonly affected organs. It was reported that more than about 70% SLE patients developed lupus nephritis, which was highly associated with the long-term prognosis1,2. It will be a great advantage if the high-risk groups could be predicated and prevented with pre-treatment, the renal prognosis and survival would be promisingly improved.

The incidence of lupus nephritis within past 10 years in new-onset SLE patients was recorded in our retrospective study, which was highest in their first-year, about 17%, and about 5% per year in the following years3. The raising of risk prediction models and the recognition of high-risk patients are quite important. The prediction model depends on the collection of patient phenotypes, which are scattered in various forms and very cumbersome. In our previous study, a total of 14,439 SLE patients were collected from the rheumatology and immunology departments of 13 Chinese tertiary hospitals in this study, including 13 062 females (90.46%), with an average age of 33.4 years, and the time span of EMR (Electronic Medical Records) was from October 28, 2001 to March 31, 2017. It includes basic information about patients, physical examination, inspection and diagnostic information, etc. We designed a hybrid NLP system combined NLP technical and expert knowledge at the same time, which was named as Deep Phenotyping System (DPS), to extract all the phenotypic information recorded in EMR.

The DPS efficiently processed EMR data, and its accuracy, precision, and recall were each greater than 93%. It extracted 73 794 entities from 14,439 SLE cases, each with time attributes, and produced 18,785,000,640 entities. Thus, a LN prediction model was raised, which the likelihood of lupus patients without nephritis will develop lupus nephritis within half and one year can be predicted.) More than 35 000 phenotypes were used in this model and it was verified with independent samples. The best accuracy (ACC) and area under the curve (AUC) predicting the 1-year and 2-year risk of developing lupus nephritis can be achieved 0.88 and 0.86 respectively.

The comprehensive SLE phenotype database constructed by NLP greatly improves the research efficiency of lupus clinical phenotype. We first proposed a predictive model of lupus nephritis, which is high applicability and efficiency. The experimental results of good close and open testing fully demonstrate the authenticity and practicality of this database. The research process and method based on real world data are also applicable to predict other important complications of lupus3.

Till now, there were no studies investigating secondary prevention tools of lupus nephritis. However, as we all known, disease flare is a high-risk factor of cruel organ damage, and our previous data showed that lupus nephritis was one of the important flare patterns4. Two phase III, randomized, placebo-controlled studies, BLISS-52 and BLISS-76 showed that belimumab, the only FDA-approved biologic in SLE, targeting B Lymphocyte Stimulator, can reduce disease flares compared to standard-of-care (SOC) therapy5,6. A propensity-score matching study further proved that belimumab add on reduces organ damage progression as measured by SDI7. A pooled post-hoc analysis of the BLISS trials took a deeper look at renal outcome, and suggest that belimumab may offer renal benefit in patients with SLE, indicated by less renal flares in belimumab group, that is 1.1% versus 3.0 in the placebo8.

We hypothesized and tried to analyze that whether belimumab could act as a secondary prevention tool for SLE patients at high-risk.

Inclusion Criteria:

• Aged ≥ 18 years old;
• Meet the 1997 revised American College of Rheumatology criteria for Systemic Lupus Erythematosus;
• 2-year risk of developing new-onset lupus nephritis higher than 50% (50%~100%) based on our AI-constructed prediction model.
• Patients received a stable SOC regimen with fixed doses of prednisone (0-40mg/day), and/or hydroxychloroquine, and/or immunosuppressive agents (azathioprine, mycophenolate, methotrexate, ciclosporin, tacrolimus, leflunomide) for at least 30 days before the study.
• Be willing to join the trial and sign the written informed consent.
• Contraception is required during the study.

Exclusion Criteria:

-- Patients with previous kidney involvement were excluded;

Definition of previous kidney involvement:

• Kidney biopsy confirmed; or
• Active urinary sediment defined as any one of the following:

A. >5 RBC/hpf in the absence of menses and infection; B. >5 White blood cell per high powered field (WBC/hpf) in the absence of infection; or Cellular casts limited to RBC or WBC casts.

C. 24-hour urine protein > 0.5g

• Severe active lupus such as neuropsychiatric lupus or cardiac involvement were excluded from the trial;

• Pregnancy or lactation;

• Previous treatment with any B-lymphocyte-targeted drug (including rituximab), intravenous cyclophosphamide within 6 months of enrolment, and intravenous Ig or prednisone (>100 mg/day) within 3 months.

• Hepatic (ALT or AST > 2 times upper normal limits) or renal dysfunction (creatinine clearance rate < 60 ml/min).

• Have a history of a primary immunodeficiency

• Have a significant IgG deficiency (IgG level < 400 mg/dL)

• Have an IgA deficiency (IgA level < 10 mg/dL)

• Infection history:

  • Currently on any suppressive therapy for a chronic infection (such as tuberculosis, pneumocystis, cytomegalovirus, herpes simplex virus, herpes zoster and atypical mycobacteria)
  • Hospitalization for treatment of infection within 60 days of Day 0.
  • Use of parenteral (IV or IM) antibiotics (anti-bacterial, antiviral, anti-Fungal, or anti-parasitic agents) within 60 days of Day 0
  • Serologic evidence of current or past Hepatitis B (HB) infection based on the results of testing for HBsAg and HBcAb
  • Have a history of an anaphylactic reaction to parenteral administration of contrast agents, human or murine proteins or monoclonal antibodies
  • Have a historically positive HIV test or test positive at screening for HIV
  • Have any other clinically significant abnormal laboratory value in the opinion of the investigator
  • Have any intercurrent significant medical or psychiatric illness that the investigator considers would make the candidate unsuitable for the study
  • Have current drug or alcohol abuse or dependence, or a history of drug or alcohol abuse or dependence within 365 days prior to Day 0

Excluded Concomitant Medications

• Anti-B-cell therapy:
• Wash-out of 5 therapeutic half lives after prior B-cell therapy, or until pharmacodynamic effect would be minimal (e.g., 1 year following rituximab)
• 365 days prior to belimumab:
• Any biologic investigational agent (e.g., abetimus sodium, anti CD40L antibody, BG9588/ IDEC 131)
• 90 days prior to belimumab:
• Intravenous cyclophosphamide
• Subjects who receive CYC whose leukocyte count is <2000/m3
• 30 Days Prior to belimumab (or 5 half-lives, whichever is greater) o Any non-biologic investigational agent
• Live vaccines within 30 days prior to baseline or concurrently with belimumab

Based on previous BLISS trials, belimumab may reduce the development or flare of lupus nephritis by over 50%8. A sample size of 48 patients per group would provide the trial with 90% power at a two-sided alpha error of 0.05 to detect an effect on reducing development of lupus nephritis in high-risk SLE patients from 50% to 25% receiving belimumab within 24 months. A model of normal approximation to one sample proportion was applied. An estimated 20% withdraw or lost-to-follow up was also calculated, thus yielding the final estimated simple size of 48 patients in each group.

Historical controls were selected by propensity-score match in our SLE database with the balance of baseline prednisone, antimalarial, immunosuppressive agents and SLEDAI. Belimumab: historical controls = 1:1