Article Text

Download PDFPDF

BD-04 Classification of SLE patients based on longitudinal assessment of complement component 3 in relation to complement component 4
  1. David R Fernandez,
  2. Mikhail Olferiev,
  3. Kyriakos Kirou,
  4. Dina Greenman,
  5. Kerri Merritt and
  6. Mary K Crow
  1. Mary Kirkland Center for Lupus Research, Hospital for Special Surgery, New York, NY, USA


Background Systemic lupus erythematosus (SLE) disease activity is characterized by tissue deposition of immune complexes and consumption of complement, which contribute to tissue injury. In clinical practice, it is common to encounter patients where either C3 or C4 is low in isolation, though the clinical implications of variation in C3 relative to C4 are unclear. Here we performed relationship-based clustering of SLE patients based on serum C3 and C4 levels to investigate if this could define distinct clinical subgroups of SLE patients.

Methods C3, C4 and other clinical and laboratory parameters were obtained from our proprietary database. A total of 151 SLE patients having an average of 38 (range 7–117) measurements of C3 or C4 were studied. To classify SLE patients based on the character of the relationship of C3 vs C4, we performed relationship-based clustering approach by defining linear fit parameters (including alpha, beta, standard error, and p values) followed by hierarchical clustering. The clusters obtained were screened in terms of their dependency to clinical data using Chi square test or Fisher’s exact test, as appropriate, with significance defined as p<0.05.

Results Clustering based on multiple characteristics of the relationship between C3 and C4 identified 6 clusters of patients. Clusters 1 and 6 were small and did not have clear phenotypes. Cluster 2 and cluster 5 were both defined by strong correlations between C3 and C4 (Cluster 2 – r=0.81, p<0.00001, Cluster 5 – r=0.81, p=0.0016), though cluster 5 had a lower median C3 level (Cluster 2 C3=79.5, Cluster 5 C3=74.5). Cluster 3 had higher median levels of C3 and C4 (C3=106.0, C4=20.6), and the correlation between C3 and C4 was far less robust (r=0.60, p=0.44730). Cluster 4 was notable for the lowest median C3 and C4 levels (C3=69.8, C4=12.3), and no significant correlation between C3 and C4 was present (r=0.54, p=0.121143).

Individuals in cluster 2 were more likely to have Jaccoud arthropathy (OR 6.11, CI 1.59 to 24.47), or a history of avascular necrosis (AVN) (OR 4.38, CI 1.55 to 12.34), but less likely to have thrombocytopenia (OR 0.15, CI 0 to 0.98). Cluster 5 patients were more likely to have thrombocytopenia (OR 2.78, CI 1.04 to 7.43) and less likely to have AVN (OR 0.27, CI 0.05 to 0.99).

Conclusions C3 and C4 levels vary widely in SLE patients but generally fall into a few general patterns, which are associated with different clinical manifestations, and may provide novel insight into underlying biological differences between SLE patients.

Acknowledgements This work was supported by the Emerald Foundation, the Barbara Volcker Center, and the Lupus and APS Center of Excellence of the Hospital for Special Surgery.

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.