Article Text

THU0161 Neuropsychiatric involvement and SLE: Performance of a new algorithm for attribution of NP events tested on an italian multicenter cohort
  1. A. Bortoluzzi1,
  2. S. Bombardieri2,
  3. C. Casu3,
  4. F. Conti4,
  5. S. De Vita5,
  6. A. Doria6,
  7. I. Farina1,
  8. G. Ferraccioli7,
  9. E. Gremese7,
  10. E. Mansutti5,
  11. M. Mosca2,
  12. M. Padovan1,
  13. M. Piga8,
  14. A. Tincani3,
  15. P. Tomietto5,
  16. C. Tani2,
  17. G. Valesini4,
  18. M. Zen6,
  19. A. Mathieu8,
  20. M. Govoni1
  21. on behalf of the Italian Study Group for the Neuropsychiatric Involvement in Systemic Lupus Erythematosus
  1. 1Rheumatology Unit, University of Ferrara, S. Anna Hospital, Ferrara
  2. 2Rheumatology Unit, University of Pisa, Pisa
  3. 3Rheumatology and Clinical Immunology Unit, University of Brescia, Brescia
  4. 4Rheumatology, University of Rome (La Sapienza), Roma
  5. 5Rheumatology Clinic, University of Udine, Udine
  6. 6Rheumatology, University of Padova, Padova
  7. 7Rheumatology, Catholic University of Rome (Sacro Cuore), Roma
  8. 8Rheumatology Unit, University of Cagliari, Cagliari, Italy


Background Neuropsychiatric (NP) involvement in systemic lupus erythematosus (SLE) includes a wide variety of neurologic and psychiatric manifestations whose attribution to the underlying disease is a clinical challenge.

Objectives To determine the sensitivity and specificity of an attribution model (AM) applied to a large multicenter series of NP events.

Methods All NP events satisfying the 1999 ACR case definition criteria occurred in a multicenter cohort of SLE patients (pts), in a timeframe of 5 years, were retrospectively analyzed. All NP events were tested by applying an arithmetical AM, firstly developed in a cohort of pts recruited in Ferrara, which included a set of 4 items selected on the basis of a literature review and by a Delphi consensus methodology during the XLVII Congress of the Italian Society of Rheumatology. The items included in the AM were the following: time of onset of NP event respect to the SLE clinical onset (before -0.5; after +0.5, concomitant +1); “minor” or not specific NP events as defined by Ainiala et al. (yes = -1; not = +0.5); opposing factors or “associations” as defined by ACR glossary (none =0; 1 = -0.5; >1 = -1); other favoring factors such as age of onset, absence of familial history for epilepsy or psychiatric disorders, abnormal serology, neuroimaging abnormalities, response to treatment, high disease activity at the time of the event (none =0, 1 =+0.5, >1=+1). According to the final score each NP event was then classified as related (≥+1), uncertain (-0.5 to +0.5) or SLE-unrelated (≤-1). The performance of the AM was estimated by its application to an external cohort of pts by the ROC curves analysis, assuming the “clinical judgment” performed by each attending team as the referral gold standard.

Results 211 eligible pts were included, 91% F e 9% M, with a mean age of 33.5 yrs for a total of 430 events evaluated. Applying the AM 3 classes of values were generated: 179 events were classified as SLE related, 29 as SLE-unrelated and the remaining as uncertain events. After exclusion of uncertain events, the result obtained by the ROC curves analysis comparing the results of the AM with the clinical judgment, yielded a sensitivity of 62% and specificity of 93% when all the NP events were considered; when the analysis was restricted to first NP event only, the sensitivity raised up to 72% and specificity was 90%.

Conclusions The overall performance of a new arithmetical AM was satisfactory in terms of sensitivity and specificity and allowed a proper placement in three-quarter of the NP events deemed as related and unrelated to SLE. This model can be regarded as an easy to use tool which allows a standardized approach to the complex sphere of NP involvement in SLE.

Disclosure of Interest None Declared

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.