Background The EULAR/ACR classification criteria for SLE have introduced three new concepts. One, ANA, with their high sensitivity, but low specificity, were re-positioned to an entry criterion. Two, all criteria items are now attributed specific weights, and this often means that two criteria, such as arthritis and antibodies to double-stranded DNA (dsDNA), are sufficient to reach the 10 point classification line. Three, instead of long lists of exclusion criteria, there is one attribution rule for all items, namely that criteria are only to be counted for SLE if there is no more likely alternative cause - such as rheumatoid arthritis for arthritis in an anti-CCP positive patient. All these concepts have individually led to discussions and misunderstandings.

Results Although ANA need to be positive only once ever, it is clear that this entry criterion still excludes patients who were always ANA negative. ANA negative SLE exists, but is uncommon, which makes this much less of a problem for classification than for diagnosis, and it is important to re-iterate that the diagnosis must be possible independent of classification criteria. In poor countries, ANA testing may be a problem, but worldwide availability of such a test must be an obvious goal. Test quality has also been discussed. Critically, not all ANA substrates had adequate sensitivity, and this has to be resolved.

For the lupus specific antibodies to Sm and dsDNA, with a relative weight of 6 points, specificity is the more critical issue. Many anti-dsDNA tests have insufficient test specificity, so that anti-dsDNA can only be counted if positive in a test with at least 90% specificity against appropriate disease controls. This is typically true for Crithidia and Farr assays, but not for many of the high throughput tests of today. Clinicians will have to be aware of the test characteristics of the anti-dsDNA tests they have available. SLE classification needs knowledge of autoantibody testing, and appropriate training of physicians will be key.

The attribution rule probably is the most challenging concept. In essence, criteria should not be counted for the classification of SLE if not accepted for expert diagnosis. In this sense, attribution is very close to the clinical process of diagnosis, as hoped for, but this needs knowledge on typical findings of other autoimmune diseases that are relevant differential diagnoses to SLE, such as primary anti-phospholipid syndrome or Sjögren’s syndrome. Appropriate attribution may be troublesome in data bases, and not following the attribution rule with diligence will automatically diminish the specificity of the new criteria. Above attribution, a misdiagnosis of for example mucocutaneous items, such as mistaking Rosacea for malar rash, will also cause trouble. Again, appropriate training will be necessary.

Despite these potential pitfalls, we believe that the EULAR/ACR criteria are a relevant step forward in appropriately defining who has SLE, and in teaching doctor and medical students how to approach a patient with possible SLE. The data have clearly demonstrated that ANA negative SLE is uncommon, and ANA are a useful entry criterion or screening parameter in case of suspected SLE. The analysis of interaction has important implications in that it has shown interactions within domains, upholding this concept, but not found significant associations between domains (or items in various domains). The latter in fact is an argument that SLE is indeed a disease, not a syndrome, and that it is the effector arm of the autoantibodies in any given SLE patient that underlies the variability.

Conclusions We will need more knowledge on autoantibodies, not less, and probably more clinical training, but this is more of a chance than a challenge. Above all, it has been remarkably easy to work together in this huge group, over the Atlantic and beyond, and I am deeply gratefully for the contribution of so many colleagues. This large team experience of collegiality and friendship will hopefully help the further worldwide collaboration that is necessary for advancing the field.

Acknowledgements The EULAR/ACR SLE classification project was equally supported by EULAR and ACR. Please refer to the publications for contributors to this effort.